Keystone: reaching-defs is the third DDG prov value

[rahlk]

Problem

skills/designing-cldk-changes/references/canonical-schema.md documents the DDG prov value as ["ssa"] (syntactic, L3) or ["points-to"] (semantic, L4). Both codeanalyzer-python (since 1.1.1, #115) and codeanalyzer-typescript also emit ["reaching-defs"] on the L4 edges that wire statements to the parameter vertices (def → actual_in, actual_out → call site, formal_in → use, def → formal_out). Without those the statement-level DDG and the port lattice are two disconnected graphs. A consumer filtering on the documented vocabulary silently drops every one of them; codeanalyzer-python#178 was filed from exactly that reading.

Scope boundary

The keystone doc only. The analyzers already emit and document it locally (codeanalyzer-python README and analyzing-canpy-graphs vocabulary as of 1.4.1).

Goals

• prov row in the edge table lists ssa, reaching-defs, points-to
• One sentence on where reaching-defs sits: syntactic side of the ladder, L4 only, port wiring
• L3/L4 DDG paragraph mentions that L4 adds both reaching-defs and points-to edges

Caveats and known risks

• None; docs only.

Definition of done

• The three prov values and their level appear in the keystone, and the monotonicity note still reads correctly (L4 widens the set).