Add correlator-independent similarity score to Version Tracking match table #9106
Description
Is your feature request related to a problem?
Follow-up to #5859. Version Tracking correlator scores reflect each correlator's matching algorithm, not actual structural similarity between matched functions. The Symbol Name correlator assigns 1.0 when names match even if functions differ significantly. For patch-diffing workflows, there is no way to sort or filter matches by how structurally similar the functions actually are.
Describe the solution you'd like
A new PDiff similarity score column computed from per-basic-block mnemonic hash comparison, stored in the match DB at creation time. Includes a score range filter and a best-match deduplication filter for cross-correlator workflows.
Additional context
I have a working implementation ready as a PR. The score uses 95% basic-block mnemonic hash similarity + 5% stack frame size similarity, persisted in a new schema v1 column with auto-upgrade from v0 sessions.
