Data.hexEncodedString() built each byte via String(format: "%02x", _),
paying Foundation format-string parsing per byte on the hot BLE receive
path (PeerID(hexData:) is called several times per received packet).
Replace it with an ASCII lookup table writing into a preallocated
buffer; output is byte-for-byte identical (lowercase hex).
Data(hexString:) similarly allocated a two-character substring and went
through integer radix parsing per byte; replace with direct nibble
lookups over the UTF-8 bytes. Semantics are unchanged except that
sign-prefixed pairs like "+f", which the old radix parser incorrectly
accepted, are now rejected.
Add round-trip, known-vector, and invalid-input tests to
BitFoundationTests.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
