feat(codec): add Zeroed type

[ygd58]

What

Adds Zeroed, a wrapper representing exactly n zero bytes — closes #1703.

Useful for padding or reserving space in an encoding, and doubles as an assertion that a reserved region was actually left untouched: decoding rejects any input where one of the n bytes is non-zero.

n is not stored on the wire; the reader supplies it externally via Read::Cfg = usize, the same way a fixed-length array's length is part of its type rather than its encoding.

Implementation

Implements Write/EncodeSize/Read following the same shape as Bytes (codec/src/types/bytes.rs), minus the length prefix (n comes from Cfg here rather than being self-describing). write() uses BufMut::put_bytes rather than allocating a temporary zero buffer.

Testing

Unit tests for round-tripping (n = 0, 1, 8, 300), rejecting a non-zero byte, rejecting a short buffer, and is_empty().

Added the arbitrary-gated conformance test block following the Bytes pattern, and generated the baseline:

just test -p commonware-codec zeroed
# 46 passed, 0 failed

RUSTFLAGS="--cfg generate_conformance_tests" just regenerate-conformance -p commonware-codec
# adds the CodecConformance<Zeroed> entry to codec/conformance.toml

Both commits included — the type itself, and the generated conformance baseline as a separate commit so the hash's provenance is clear.

Closes #1703

[cursor]

Cursor Bugbot has reviewed your changes using high effort and found 3 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

Want fixes drafted automatically? Bugbot Autofix can create code changes for findings. A team admin can enable Autofix in the Cursor dashboard.

^{Reviewed by Cursor Bugbot for commit 4b3319a. Configure here.}

[ygd58]

Addressed all 3 Bugbot findings in the follow-up commit:

1. Decode reallocates for zero check — read_cfg now uses util::ensure_zeros instead of copying into a Vec and scanning it.
2. Pooled encode double-allocates zeros — overrode encode_inline_size to return 0, since write_bufs pushes the zero region as a separate chunk rather than writing it inline.
3. Missing write_bufs equivalence test — added test_zeroed_write_bufs_matches_write using the existing TrackingWriteBuf test double, asserting write and write_bufs produce identical bytes.

write() itself is unchanged, so the conformance baseline from the prior commit still applies.

Verified on a real toolchain (rustc >= 1.95, which I don't have access to directly):

just test -p commonware-codec zeroed
# 6 tests run: 6 passed, 128 skipped