pub fn classify_algorithm_names(name: &str) -> Vec<AlgorithmClassification>Expand description
Extract every recognizable algorithm mention from a free-form name using
word-boundary token matching (never bare substrings). Used for cipher-suite
names (“TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256”) and as the guarded
name-based fallback for assets without algorithmFamily/OID.
Tokens are split on any non-alphanumeric character (keeping + for
“SPHINCS+”); spans of up to three adjacent tokens are joined with - and
matched longest-first, so “ML KEM 1024” and “AES 128” resolve as units.
Unrecognized tokens (GCM, CBC, TLS, WITH, …) are simply skipped.