* talker_alias.py: replace dmr_utils3.bptc.encode_emblc with a spec-correct
in-module _encode_emblc. The library version has a transcription bug in burst
D (uses _binlc[24] twice instead of _binlc[25]) that flips one bit of byte 2
of every embedded LC. Harmless for the group voice LC (that bit lands in
Service Options) but for Talker Alias byte 2 is the header (format/length) of
block 0 and the first text char of blocks 1-3, so it corrupts the alias and
consumes the embedded-LC error-correction budget. Validated bit-for-bit
against the known-good real DMR bursts shipped in dmr_utils3.decode.
* talker_alias.py: add load_ta_profiles() and expose {city}/{state}/{country}
to TALKER_ALIAS_FORMAT.
* bridge_master.py: build _TA_PROFILES from the full subscriber file (callsign +
fname/surname/city/...) with callsign-only fallback, so inject-mode templates
like "{callsign} {fname} {city}" work (previously only {callsign} was
populated, leaving {fname} empty).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adan C31AG
Joaquin Madrid Belando