Fallback to UniDic aType-derived lexical pitch when OJAD is unavailable

[torrid-fish]

Problem

UniDic's per-morpheme accent kernel (aType) is already tokenised and carried on every word as lexical_kernel, but it does not drive accent_marking_type: when no connected-speech contour is available a token degrades to flat pitch (accent_marking_type=0) instead of an approximate dictionary-form contour. Use lexical_kernel to synthesise a per-mora contour (Tokyo-dialect 平板 / 頭高 / 中高・尾高 mapping) as a floor, so degraded output carries an approximate accent rather than nothing.

[torrid-fish]

Tracked under the standalone epic #62 as the lexical-pitch floor tier (used when no prosody engine / OJAD is available).

[torrid-fish]

Closing because this problem no longer exists, not because the proposed feature was built. PR #63 removes OJAD entirely, so there is no "OJAD-down" state to fall back from: the connected-speech contour now comes from in-process OpenJTalk, and UniDic aType (lexical_kernel) is already consumed inside api/accent/openjtalk.py. The lexical-pitch floor this issue asked for is therefore moot. Tracked historically under epic #62.