docs: consumer prose about the subc tree has no falsifier on a path dependency (#122)

[iceteaSA]

Linked issue

Approved issue: #122

What changes

Two rows appended to the numbered table in docs/hunting-loop-briefing.md, as 557 and 558. I appended them rather than inserting them next to 278, because the file refers to rows by number (for example "row 324") and inserting would renumber everything after 278. Row 557 names 278 as its companion.

• 557: a consumer's present-tense record of a path dependency's tree (a version, a capability, "not exported yet") has no falsifier and decays silently. It carries your narrowing: a version-pinned crates.io dependency has one in its version requirement, so the row applies to the remaining path consumers and to capability claims, which no version requirement catches. The tense distinction is kept as written: dated history is fine, present tense needs a mechanical check.
• 558: prefer reading SUBC_PROTOCOL_CRATE_VERSION, or calling build_provenance, over writing the value down in prose.

And one sentence in docs/fleet-surface.md under Provenance reads: a consumer's own documentation of which subc version it runs isn't provenance, the supervisor.provenance declaration is, with a pointer to the two rows.

Docs only. git diff --check is clean; the repo has no markdown lint to run.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

---

Summary by cubic

Adds prose clarifying that a consumer's present-tense record of a path dependency's tree has no falsifier and decays silently, and that reading SUBC_PROTOCOL_CRATE_VERSION or calling build_provenance is preferable to writing values down. Addresses #122.

• Appends rows 557 and 558 to the table in docs/hunting-loop-briefing.md rather than inserting near row 278, so existing row-number references stay valid.
• Adds a sentence to docs/fleet-surface.md noting that a consumer's own documentation of which subc version it runs isn't provenance; the supervisor.provenance declaration is.

^{Written for commit 7312908. Summary will update on new commits.}

[cortexkit-ci]

#122 has the design-approved label; this PR can be reviewed.

[iceteaSA]

The red Test (windows) on the previous head wasn't this PR: ck-bus --test signer_shape :: production_binary_has_no_path_to_the_harness_signer failed on the merge ref with d3a35346, which was red on the same test on master and fixed by 058a864c. #120 and #127 ran after the fix and are green. I rebased onto 67f40969 (the range-diff is identical, ebb9f59b = 73129088) so CI re-runs with the fix.