refactor(#57): rename utils→b64, split into int/binary/charset, + no_std stream fix

[joeldsouzax]

Phases 1a–1b of the module/naming restructure (#57), plus a folded-in no_std CI fix. Behaviour-preserving throughout — gate green including keripy vectors, cesr-wasm, and cesr-nostd.

What changed

Phase 1a — utils → b64 (9eb3b96): the utils module was a non-name for the CESR Base64 codec. Renamed module + cargo feature utils → b64, and killed the utils::utils module-inception (→ b64::alphabet). Whole-crate paths, lib.rs, Cargo.toml, flake.nix, and the frozen_surface API-pin test updated.

Phase 1b — split by responsibility (26e2bfd): b64/encode.rs + b64/decode.rs → b64/int.rs (int↔b64, both directions) + b64/binary.rs (bytes→b64); extracted b64/charset.rs (is_b64_url_safe_charset) out of b64/alphabet.rs. b64/ is now alphabet · int · binary · charset · error.

no_std CI fix (ac513f4): cesr-nostd built alloc,core,b64,keri — omitting stream, even though stream is no_std-clean. Added it so the no_std check exercises the stream parser. (crypto/serder stay out: getrandom has no bare-target backend.)

⚠️ Breaking (public paths)

• cesr::utils::* → cesr::b64::*
• cargo feature utils → b64
• submodule paths cesr::b64::encode::* / ::decode::* → cesr::b64::int::* / ::binary::*

The re-exports (cesr::b64::encode_int, decode_to_int, encode_binary, is_b64_url_safe_charset) are stable. 0.x → MINOR bump.

Not in this PR (follow-up #57 phases)

Fold stream/util into b64; rename core/utils → core/code_size; naming-convention sweep + CLAUDE.md update; CI trigger dedup. Tracked on #57.

Verification

nix flake check green: clippy, all tests + keripy differential vectors, doctest, cesr-wasm, cesr-nostd (now with stream).

Refs #57

🤖 Generated with Claude Code

[codspeed-hq]

Merging this PR will not alter performance

✅ 11 untouched benchmarks

---

_{Comparing refactor/57-b64-module (2066545) with main (07a75d4)}