dump-catalog: expose per-domain terms + extractor functions

[coderdan]

Follow-up from #364 (the structured eql-manifest.json).

eql-codegen dump-catalog currently emits, per domain: segment / suffix / supported_ops. The Term enum (crates/eql-domains/src/term.rs) already knows each term's:

• json_key() — hm / ob / bf / op
• extractor() — eq_term / ord_term / match_term / ord_ope_term
• ctor() — hmac_256 / ore_block_256 / bloom_filter / ope_cllw

Exposing the per-domain terms (with those names) in DomainEntry would:

• link each domain to its extractor functions authoritatively in the manifest;
• resolve false-positives in the docs drift-lint (docs(v2): consume the EQL manifest — generated function catalog + drift-lint docs#46), which flags eq_term / ord_term / match_term — these are real generated extractors, but doxygen doesn't capture them cleanly, so the catalog is the authoritative source for the names.

Change: add a terms: Vec<TermInfo> (json_key + extractor + ctor) to DomainEntry in crates/eql-codegen/src/dump.rs, populated from d.terms + the Term accessors — a few lines; the data is all there. The manifest generator then maps it through.

Context: PR #364, cipherstash/docs#46.