docs: update outdated stats across all documentation - #774

Maintainer edits available on all branches; CI green where run; nullclaw#777/nullclaw#775/nullclaw#774
conflict (branch updates required); all PRs BLOCKED by review-required
protection — first merge will test whether maintain role clears it. donprus
courtesy done via Discord by user.

Supersedes nullclaw#774. Every figure below was recomputed from source at
main (5f1cade) rather than carried forward, because nullclaw#774's own numbers
had themselves gone stale: it proposed 6,300+ tests against an actual
7,499, and ~2.7 MB for a binary that measures 4.66 MB.

## Scale figures

- Source files 245 -> 293; src LOC ~204K -> ~290K; tests 5,640+ -> 7,499.
- Providers: "50+ implementations (9 core + 41 compatible services)" ->
  10 core implementations plus 110 OpenAI-compatible registry entries,
  counted from the `.{ .name = ... }` table in `src/providers/factory.zig`,
  which is the actual single source of truth for compatible providers.
- Channels 17 -> 24; the architecture tables listed a partial set that
  omitted Teams, Max, WeChat/WeCom, and Weixin. Now labelled with the
  count and the omissions filled in. English and Chinese kept in sync.
- Tools "30+" -> 40 registered implementations, counted from
  `tool_name` constants in `src/tools/`.

## Binary size — a real discrepancy, not just a stale number

`README.md` and `CLAUDE.md` claimed 678 KB, and AGENTS.md set a sub-1 MB
ReleaseSmall target. A host build measures 4,889,528 bytes (~4.66 MB,
aarch64-macOS):

    zig build -Doptimize=ReleaseSmall && ls -l zig-out/bin/nullclaw

That is roughly 7x the long-quoted figure and about 4.7x the stated
target. All three now state the measured value, AGENTS.md marks the
sub-1 MB goal as an open gap rather than a met constraint, and a
verification command is included so the figure is re-checkable instead
of asserted.

This is a product-accuracy finding, not only a docs fix: the headline
size claim has been wrong for a long time. Worth deciding separately
whether to shrink the binary or restate the target.

## Keeping the numbers current

The recurring failure is that these figures are hand-maintained and
silently rot — nothing in CI checks them, which is also why nullclaw#774 went
stale in the first place. AGENTS.md §1 now carries the exact commands
used to derive each number, and all of them were confirmed to run as
written. A repo-wide sweep confirms no remaining 5,640 / 5,300 / 678 KB
/ 0.15.x references in the touched files.

Note for reviewers: the "last verified at 5f1cade" marker is the point
at which these numbers stop being true. Please re-derive rather than
increment them.

Docs only, no code touched. `zig build test --summary all` 13/13 steps,
7490/7499 passed, 9 skipped, 0 failures, 0 leaks. `zig build
-Doptimize=ReleaseSmall` and `zig fmt --check src/` both clean.