tests@v20.0.1

Patch BPO2 mainnet fixture release for tests@v20.0.0. This is a patch (Z) release under the tests@vX.Y.Z scheme: There are no consensus-impacting changes; this release covers test additions and other fixture modifications.

Highlights

• feat(test-fill): pack Engine X pre-alloc groups by @danceratopz in #3122

  This removes 2179 pre-alloc groups from the blockchain_test_engine_x fixtures (from 5458 to 3279). The eels/consume-enginex simulator boots one execution client per pre-alloc group and reuses it across the group's tests, so the group count is the main driver of simulator runtime: this is roughly 40% fewer client boots per run. Boot-to-ready times below are measured from the client logs of a consume-engine run (one client per test), which set the per-group cost that is saved:

  Client	Boot to ready	Removed per run (x 2,179)	At --sim.parallelism=4
  besu	~6.7 s	~4.0 h	~1.0 h
  erigon	~0.45 s	~16 min	~4 min
  ethrex	~0.08 s	~3 min	~45 s
  geth	~0.43 s	~15 min	~4 min
  nethermind	~3 s	~1.8 h	~27 min
  nimbus-el	~2.8 s	~1.7 h	~25 min
  reth	~0.83 s	~30 min	~8 min

  Boot times exclude Docker container start and stop overhead.

New Tests

• feat(tests): access-list slot warmth survives a failed CREATE2 by @chfast in #3086

• fix(tests): test tx max nonce at u64 boundary, add nonce overflow test by @spencer-tb in #3165

  Adds transaction nonce validation coverage from Frontier onwards: rejection at the U64.MAX_VALUE boundary and rejection of nonce overflow.

• feat(tests): restore sstore_combinations as a hand-written test by @chfast in #3154

  Restores sstore_combinations as a hand-written Istanbul test (EIP-2200) after it was deleted accidentally by the ported-static regeneration sweep. Twelve orphaned ported_static filler stubs are removed and the matrix is reduced from 5187 to 774 cases per fork, with each reduction step trace-verified to keep the same SSTORE coverage.

Changed and Fixed Tests

• chore(specs): reorder tx validity checks by @SamWilsn in #3036

  Reorders the transaction validity checks across the mainnet-fork specs (order only, no change to any state-transition result), and updates test_tx_nonce to emit state-test fixtures instead of blockchain-test fixtures.

Fixture Format Changes

• feat(test-fill): pack Engine X pre-alloc groups by @danceratopz in #3122

• fix(test-fixtures): emit chainId in state test fixture transaction by @spencer-tb in #3125

  State-test fixtures now emit chainId in the transaction, so client state-test consumers should expect the field on every state test.