[project-s] ハミング機能・歌機能向けのモデル・API(compatible_engineのみ)を追加 (#724)

* remove contour and rename to talk xxx

* fix speaker id map

* rename functions and variables

* add models to model file

* add sing style and source filter models to model file set

* add new models to status

* rename get model index and speaker id

* add new models session

* change i32 to i64

* add new predictor to inference core

* add new predictor to core

* add new predictor to compatible engine

* rename source filter to sf decode

* fix rename miss

* rename sing style to sing teacher

* fix rename miss

* remove vector

* add TODO comment (add sing tests)

Co-authored-by: Hiroshiba <[email protected]>

* fix comment out

* lint

---------

Co-authored-by: Hiroshiba <[email protected]>