Refactor E2E suites to allow AHM integration

[rockbmb]

Closes #353, closes #382, closes #258.

Regarding #353

In order to keep a single test suite that can, with minimal changes, be run both on

• live relay chain runtimes, and also
• zombie-bite forks of post-migration Asset Hubs (used for testing)

some changes are necessary, most importantly to the scheduler E2E test suite.
This pallet, in a post-migration Asset Hub runtime setting, will use a non-local block provider, which changes some of the expectations that PET E2E tests have; see this PR for an example change.

Regarding #382

Furthermore, this PR also changes some dev accounts used in testing to fresh accounts with no state - Westend and Paseo runs of PET had false positives due to unclean states for Alice, Bob, etc.

Regarding #258

With the changes required by #353, the scheduler E2E test suite now has tests that are expected to run on relay chains, parachains with local block providers, and parachains with non-local block providers.
This entailed several .match() pattern matches to determine the correct offset at which to schedule tasks.

Documentation on the module and each test has been improved to clarify the new branching logic of these tests.

---

• governance
• multisig
• nomination pools
• staking
• vesting
• scheduler
• proxy

[github-actions]

The changes are a significant and valuable refactoring to enable E2E tests on Asset Hub runtimes, primarily by abstracting away relay vs. para chain differences and using fresh accounts to prevent state conflicts. The addition of await to async calls fixes potential race conditions. However, there are several opportunities to improve clarity and robustness in the staking tests.

[xlc]

can you merge master
and feel free to resolve/ignore AI review comments. but sometimes they are useful.

[rockbmb]

CI is failing due to the introduction of a paseo package for local testing.

I will remove this before completing this PR.

[rockbmb]

Kusama scheduler E2E tests are failing.

The behavior of scheduler.incompleteSince has changed on Kusama; if related to #397, I create a PR to fix tests/snapshots and then merge master into this branch.

[rockbmb]

@xlc I still need to update scheduler tests for Kusama, but this is ready to be looked at.

In order to update proxy E2E tests for use with post-AHM parachains, I had to update them in the same manner you did with #399.
Polkadot relay proxy E2E tests are still failing for me, but after a fix, #398 can be closed in favor of this, if you're alright with that.

[github-actions]

The changes introduce a new test configuration to support different block providers, which is a great improvement for testing across different chain types. However, I found one potential issue in the staking tests where a performance optimization is now only applied to local block providers, which might affect tests on other chains.

---

Review Suggestions

packages/shared/src/staking.ts:1948-1963

The logic to remove storage prefixes for ParasDisputes, Dmp, Staking, and Session is now only executed for 'Local' block providers. This might cause performance issues or flakiness in tests for non-local block providers, such as the Asset Hubs targeted by this PR. If this optimization is still needed, it should be applied to all block providers or have a clear justification for why it's not needed for non-local ones.

[rockbmb]

The two failing tests belong to Kusama's scheduler E2E test suite.

They are:

1. test that schedules an excessively overweight task, which should not be executed

   • polkadot-ecosystem-tests/packages/shared/src/scheduler.ts

     Lines 559 to 570 in 7879220

     	/**
     	* Test the process of
     	*
     	* 1. creating a call requiring a `Root` origin: an update to total issuance
     	* 2. artificially manipulating that call's weight to the per-block weight limit allotted to scheduled calls
     	* 3. scheduling the call
     	* 4. checking that the call was not executed
     	* 5. checking that it remains in the agenda for the original block it was scheduled in
     	* @param chain
     	*
     	*/
     	export async function scheduledOverweightCallFails<

2. test that schedules two differently-prioritized tasks, heavy tasks in the same block, and observes that priority is respected

   • polkadot-ecosystem-tests/packages/shared/src/scheduler.ts

     Lines 559 to 570 in 7879220

     	/**
     	* Test the process of
     	*
     	* 1. creating a call requiring a `Root` origin: an update to total issuance
     	* 2. artificially manipulating that call's weight to the per-block weight limit allotted to scheduled calls
     	* 3. scheduling the call
     	* 4. checking that the call was not executed
     	* 5. checking that it remains in the agenda for the original block it was scheduled in
     	* @param chain
     	*
     	*/
     	export async function scheduledOverweightCallFails<

The first one is failing because on Kusama, scheduler.incompleteSince remains set after attempting to execute the overweight task, as expected.
The second fails because even after executing the lower-priority task, scheduler.incompleteSince remains set - this is not what should happen.

I will push a commit to allow different behavior on Kusama for now; investigating this is outside the scope of this PR, which has grown large enough.