Correct OuteTTS model_specs_v1 metadata

[mirek190]

What

Follow-up to #63 that brings the OuteTTS catalog entry in model_specs_v1/ in line with the current loader and package layout.

• Rename dac_graph_arena_mb to the runtime-consumed dac_graph_context_mb.
• Rename aligner_path to the runtime-consumed aligner_model_path.
• Point the package target at the published/tested Llama-OuteTTS-1.0-1B_Q8 directory.
• Document that the standalone GGUF embeds the IBM DAC codec and Qwen3 Forced Aligner assets needed for one-file TTS and cloning.

Why

The previous option names were metadata-only aliases that the OuteTTS loader does not consume, so settings produced from the v1 catalog would be silently ignored. The old target directory also differed from the package layout used by the published model and existing examples.

Impact

This is a catalog-only correction. It does not change OuteTTS inference code or model outputs. Clients consuming model_specs_v1/outetts.json now emit the exact option keys accepted by the loader and install the package into the expected directory.

The existing 23-language list, tasks, capabilities, source declarations, and runtime bridge remain unchanged.

Checks

• python -m json.tool model_specs_v1/outetts.json
• Confirmed v1 sources exactly match model_specs/outetts.json
• Confirmed all current request and session options against src/community_models/outetts/loader.cpp
• Confirmed 23 language entries
• python tools/model_manager_v2.py --specs-dir model_specs_v1 info outetts
• python tools/model_manager_v2.py --specs-dir model_specs_v1 install outetts --models-root ../models --check
  • resolved Text to audio (TTS)/Llama-OuteTTS-1.0-1B_Q8.gguf
  • remote size: 3,029,895,456 bytes
• git diff --check

[mirek190]

Follow-up verification for the OuteTTS spec migration:

Previous v1 metadata	Current loader key	Result
dac_graph_arena_mb	outetts.dac_graph_context_mb	corrected to dac_graph_context_mb
aligner_path	outetts.aligner_model_path	corrected to aligner_model_path
Llama-OuteTTS-1.0-1B-GGUF	published package directory	corrected to Llama-OuteTTS-1.0-1B_Q8

Additional checks:

• The v1 source declarations remain byte-for-byte equivalent to the runtime bridge in model_specs/outetts.json.
• All 12 request options and all 8 session options were checked against the current OuteTTS loader.
• The existing 23-language metadata is retained.
• The model-manager package check resolves Text to audio (TTS)/Llama-OuteTTS-1.0-1B_Q8.gguf at 3,029,895,456 bytes.
• The package description now makes the standalone layout explicit: OuteTTS weights plus embedded IBM DAC and Qwen3 Forced Aligner resources, so the GGUF supports one-file synthesis and cloning.
• This PR changes catalog metadata only; inference code and output behavior are unchanged.