Column locator sequence input

[ChristianGeng]

Goal

close #407: Typing provided squence type input to Dependencies._column_loc but it wasn't implemented.

Solution

There was a mismatch between the type hints suggesting that it is possible to use sequence input and output.
We have chosen to update the type hints as the suggested vectorized implenentation did not provide large benefits.

Therefore this corrects the type hints allowing only scalar in- and putput.

[hagenw]

Thanks for trying to tackle this.

The wrong typing most likely comes from my test supporting lists instead of scalars to speed processing up in #370, but as I was not able to find a meaningful speed up, I reverted the changes in #393.

I would suggest to solve #407 by changing the typing to support only scalars.

[hagenw]

Thanks for updating the code. Can you also update the description of the pull request, so it's easy to see later on, what was added here.

[hagenw]

BTW, I checked how we avoid calling [deps.bit_depth(file) for file in files] when we want to get the information for all files in audb.info.bit_depth(), and the trick is to just get the column from the dependency table:

audb/audb/core/info.py

Line 109 in 075b279

return set(df[df.type == define.DependType.MEDIA].bit_depth)

[hagenw]

As stated in #415 (comment) this is ready to merge, but would be nice to first update the description.

[ChristianGeng]

As stated in #415 (comment) this is ready to merge, but would be nice to first update the description.

I have updated the PR description above. Merging.