Ensure outbound XCMs are decodable with limits + add `EnsureDecodableXcm` router (for testing purposes) (#4186)

This PR:
- adds `EnsureDecodableXcm` (testing) router that attempts to *encode*
and *decode* passed XCM `message` to ensure that the receiving side will
be able to decode, at least with the same XCM version.
- fixes `pallet_xcm` / `pallet_xcm_benchmarks` assets data generation

Relates to investigation of
https://substrate.stackexchange.com/questions/11288 and missing fix
https://github.com/paritytech/polkadot-sdk/pull/2129 which did not get
into the fellows 1.1.X release.

## Questions/TODOs

- [x] fix XCM benchmarks, which produces undecodable data - new router
catched at least two cases
  - `BoundedVec exceeds its limit`
  - `Fungible asset of zero amount is not allowed`  
- [x] do we need to add `sort` to the `prepend_with` as we did for
reanchor [here](https://github.com/paritytech/polkadot-sdk/pull/2129)?
@serban300 (**created separate/follow-up PR**:
https://github.com/paritytech/polkadot-sdk/pull/4235)
- [x] We added decoding check to `XcmpQueue` -> `validate_xcm_nesting`,
why not to added to the `ParentAsUmp` or `ChildParachainRouter`?
@franciscoaguirre

 (**created separate/follow-up PR**:
https://github.com/paritytech/polkadot-sdk/pull/4236)
- [ ] `SendController::send_blob` replace `VersionedXcm::<()>::decode(`
with `VersionedXcm::<()>::decode_with_depth_limit(MAX_XCM_DECODE_DEPTH,
data)` ?

---------

Co-authored-by: Adrian Catangiu <[email protected]>

## Problem
During the bumping of the `polkadot-fellows` repository to
`[email protected]`, I encountered a situation where the benchmarks
`teleport_assets` and `reserve_transfer_assets` in AssetHubKusama
started to fail. This issue arose due to a decreased ED balance for
AssetHubs introduced
[here](https://github.com/polkadot-fellows/runtimes/pull/158/files#diff-80668ff8e793b64f36a9a3ec512df5cbca4ad448c157a5d81abda1b15f35f1daR213),
and also because of a [missing CI
pipeline](https://github.com/polkadot-fellows/runtimes/issues/197) to
check the benchmarks, which went unnoticed.

These benchmarks expect the `caller` to have enough:
1. balance to transfer (BTT)
2. balance for paying delivery (BFPD).
 
So the initial balance was calculated as `ED * 100`, which seems
reasonable:
```
const ED_MULTIPLIER: u32 = 100;
let balance = existential_deposit.saturating_mul(ED_MULTIPLIER.into());`
```
The problem arises when the price for delivery is 100 times higher than
the existential deposit. In other words, when `ED * 100` does not cover
`BTT` + `BFPD`.

I check AHR/AHW/AHK/AHP and this problem has only AssetHubKusama
```
ED: 3333333
calculated price to parent delivery:  1031666634  (from xcm logs from the benchmark)
---

3333333 * 100 - BTT(3333333) - BFPD(1031666634) = −701666667
```
which results in the error;
```
2024-02-23 09:19:42 Unable to charge fee with error Module(ModuleError { index: 31, error: [17, 0, 0, 0], message: Some("FeesNotMet") })
Error: Input("Benchmark pallet_xcm::reserve_transfer_assets failed: FeesNotMet")
     
```

## Solution

The benchmarks `teleport_assets` and `reserve_transfer_assets` were
fixed by removing `ED * 100` and replacing it with `DeliveryHelper`
logic, which calculates the (almost real) price for delivery and sets it
along with the existential deposit as the initial balance for the
account used in the benchmark.


## TODO

- [ ] patch for 1.6 -
https://github.com/paritytech/polkadot-sdk/pull/3466
- [ ] patch for 1.7 -
https://github.com/paritytech/polkadot-sdk/pull/3465
- [ ] patch for 1.8 - TODO: PR

---------

Co-authored-by: Francisco Aguirre <[email protected]>

# Note for reviewer

Most changes are just syntax changes necessary for the new version.
Most important files should be the ones under the `xcm` folder.

# Description 

Added XCMv4.

## Removed `Multi` prefix
The following types have been renamed:
- MultiLocation -> Location
- MultiAsset -> Asset
- MultiAssets -> Assets
- InteriorMultiLocation -> InteriorLocation
- MultiAssetFilter -> AssetFilter
- VersionedMultiAsset -> VersionedAsset
- WildMultiAsset -> WildAsset
- VersionedMultiLocation -> VersionedLocation

In order to fix a name conflict, the `Assets` in `xcm-executor` were
renamed to `HoldingAssets`, as they represent assets in holding.

## Removed `Abstract` asset id

It was not being used anywhere and this simplifies the code.

Now assets are just constructed as follows:

```rust
let asset: Asset = (AssetId(Location::new(1, Here)), 100u128).into();
```

No need for specifying `Concrete` anymore.

## Outcome is now a named fields struct

Instead of

```rust
pub enum Outcome {
  Complete(Weight),
  Incomplete(Weight, Error),
  Error(Error),
}
```

we now have

```rust
pub enum Outcome {
  Complete { used: Weight },
  Incomplete { used: Weight, error: Error },
  Error { error: Error },
}
```

## Added Reanchorable trait

Now both locations and assets implement this trait, making it easier to
reanchor both.

## New syntax for building locations and junctions

Now junctions are built using the following methods:

```rust
let location = Location {
    parents: 1,
    interior: [Parachain(1000), PalletInstance(50), GeneralIndex(1984)].into()
};
```

or

```rust
let location = Location::new(1, [Parachain(1000), PalletInstance(50), GeneralIndex(1984)]);
```

And they are matched like so:

```rust
match location.unpack() {
  (1, [Parachain(id)]) => ...
  (0, Here) => ...,
  (1, [_]) => ...,
}
```

This syntax is mandatory in v4, and has been also implemented for v2 and
v3 for easier migration.

This was needed to make all sizes smaller.

# TODO
- [x] Scaffold v4
- [x] Port github.com/paritytech/polkadot/pull/7236
- [x] Remove `Multi` prefix
- [x] Remove `Abstract` asset id

---------

Co-authored-by: command-bot <>
Co-authored-by: Keith Yeung <[email protected]>

This was discovered during a debugging session, and it only served to
mask the underlying error, which was not great.

* Document non-uniqueness of SetTopic IDs

* More comments on WithUniqueTopic

fix for:
```
error: using `clone` on type `[u8; 32]` which implements the `Copy`
trait
  --> xcm/xcm-builder/src/routing.rs:44:28
   |
44 |             message.0.push(SetTopic(unique_id.clone()));
   |                                     ^^^^^^^^^^^^^^^^^ help: try
removing the `clone` call: `unique_id`
   |
   = help: for further information visit
https://rust-lang.github.io/rust-clippy/master/index.html#clone_on_copy
   = note: `-D clippy::clone-on-copy` implied by `-D clippy::complexity`
```

* Tools for unique topic references

* Formatting

* Naming

* Repot into routing.rs.

* More things done

* Universal Exporter supports topic-as-reference

* Some tests for the topic routing

* More tests

* Paid bridge tests

* Add message ID to sending events

* Formatting

* fix and integrate into test nets

* Move DenyThenTry and friend from Cumulus

* Append SetTopic rather than prepend

* Docs

* Docs

* Work with new ProcessMessage ID API

* Formatting

* Fix build

* Fixes

* Formatting

* Update xcm/xcm-builder/src/barriers.rs

Co-authored-by: Francisco Aguirre <[email protected]>

* Update xcm/xcm-builder/src/routing.rs

Co-authored-by: Francisco Aguirre <[email protected]>

* Docs

* Rename message_hash

* Formatting

* ".git/.scripts/commands/fmt/fmt.sh"

* Rename

* Another Rename

* ".git/.scripts/commands/fmt/fmt.sh"

* ".git/.scripts/commands/fmt/fmt.sh"

* Update xcm/xcm-builder/src/routing.rs

Co-authored-by: Keith Yeung <[email protected]>

---------

Co-authored-by: Francisco Aguirre <[email protected]>
Co-authored-by: command-bot <>
Co-authored-by: Keith Yeung <[email protected]>