execution/commitment: materialize collapse-sibling branch in toWitnessTrie

[awskii]

debug_executionWitness blinded the surviving sibling of a collapsing storage branch on a zero-value write (#21869). On a slot delete from a two-child branch, toWitnessTrie emitted the sibling extension but left the branch it points to a bare HashNode. reth's stateless verifier can't re-form the collapsed branch from a blinded node, so it returns BlindedNode and the post-state root diverges.

Repro: mainnet block 25335936, zero-write to slot 0x43cb…, sibling 0x43cd blinded → SparseTrieError(BlindedNode(Nibbles(0x43cd))).

Fix: in exclusion-proof (legacy) mode, materialize that branch via witnessMaterializeBranch with a subtrie-root check instead of a HashNode. Same hash, witness root unchanged; canonical mode stays minimal. The absent-slot path and this one now share witnessMaterializeBranchChild.

Test: Test_WitnessTrie_GenerateWitness/CollapseSiblingExtensionMustBeMaterialized — two-child storage branch, sibling is an extension over a sub-branch, delete the other child, assert the sub-branch is a FullNode not a HashNode. End-to-end: block 25335936 witness passes reth stateless (VALID; state nodes 16395 → 16396).

Fixes #21869

[awskii]

Comparing debug_executionWitness (legacy mode) output against reth on a mainnet archive node: erigon's state nodes are a strict subset of reth's (erigon_only=0), codes and keys identical, ~20–47 fewer state nodes per block. Characterizing one block's reth-only set:

What the reth-only nodes are

• 20 branches + 12 leaves
• 31 of 32 are NOT referenced by any erigon node — they live in subtrees erigon's witness never points into, i.e. they are not children erigon blinded on its own proof paths.
• Only 1 is a node erigon hash-references but omits (a genuinely blinded child).
• Leaf samples show the tell: the same key 20290decd9 appears twice with different value lengths (vlen=21 and vlen=2), and keys carry the 0x20 leaf prefix — i.e. pre- and post-state versions of modified slots. reth's witness carries touched/after-state nodes that erigon's minimal recompute witness doesn't.

Do we need them?

No. Three independent reasons:

1. erigon's witness is VALID under reth's own stateless verifier — it re-executes the block and reproduces the post-state root without those 32 nodes. If any were required, re-execution would fail (BlindedNode or wrong root). It doesn't.
2. 31/32 aren't even on erigon's proof paths — they're reth-side completeness (after-state / read-touched nodes), not data erigon dropped.
3. Across every block checked, erigon_only=0, codes and keys identical — erigon never omits a node it references; it's strictly leaner, never short.