fix(testing): tell the fixture when slot processing is skipped

[gopikannappan]

Closes #1198.

test_block_with_wrong_slot and test_block_at_parent_slot_rejected_when_slot_processing_skipped both process a block against a state deliberately left behind the block's slot, because that mismatch is the rejection they test. The spec arranges it with BlockSpec.skip_slot_processing, but the emitted fixture said nothing about it.

So a runner replaying the fixture calls the whole state transition, advances the slot first, never reaches the mismatch, and has no way to tell whether the fixture is wrong or its own client is. As the issue puts it, test runners cannot know which tests they should skip slot processing for.

The change

The fixture gains a per-block skipSlotProcessing list:

"skipSlotProcessing": [true]

It is emitted only when at least one block needs it. Every other fixture is byte-identical to before, and an absent field keeps meaning exactly what runners already do, so nothing existing has to change.

I wrote it as a list beside blocks rather than as something attached to each block, because a block is an SSZ container and this is an instruction to the runner rather than part of the block.

Checked against the real output

Not just unit tested. Running fill for the two tests named in the issue, both fixtures now carry skipSlotProcessing: [true], and test_block_extends_deep_chain, an ordinary test, emits no such field. The determinism check passes in both runs.

Three tests cover it: an ordinary run emits nothing, a skipping run emits the flag, and a run that mixes both pins [true, false]. That last one matters, because a single flag or a list built in the wrong order would pass a test where every block skips. I checked both mutations: reversing the list fails the ordering test, and never emitting the field fails two of the three.

ruff check, ruff format --check, ty check and codespell are all clean.