Relax arithmetic rules in case only one audio object is involved

[f-brinkmann]

Which issue(s) are closed by this pull request?

Closes #587

Changes proposed in this pull request:

• Bypass rules for fft norm if only one audio object is involved in the opeeration
• Add and update tests
• Update docstrings

For discussion

I discovered that we had a test that explicitly does not allow 1/signal if the FFT norm is not 'none'. I did not remove this test yet. Do we want to stay hard?

[ahms5]

Thank you for fixing! How about testing if its really working, to make sure it will work in future as well.

[f-brinkmann]

Thank you for fixing! How about testing if its really working, to make sure it will work in future as well.

I did a quick check locally, and everything works. I'll add it if we decide to change the behaviour - I added a todo in the PR description to make sure we dont forget. Just didn't want to do too much work that might be discarded :)

[ahms5]

I would suggest not to change the behaviour of _match_fft_norm instead jsut call it if we have more then 1 Audio object

[ahms5]

Suggested change

	for d in enumerate(data):
	for _, d in enumerate(data):

otherwise sth breaks

[ahms5]

Suggested change

	fft_norm = _match_fft_norm(
	fft_norm, d.fft_norm, division, n_audio_objects)
	fft_norm = d.fft_norm if n_audio_objects < 2 else \
	_match_fft_norm(fft_norm, d.fft_norm, division)

I would not change the behaviour of _match_fft_norm, instead i would just call it, if we have more then 1 audio object.

[ahms5]

Suggested change

	def _match_fft_norm(fft_norm_1, fft_norm_2, division=False, n_audio_objects=2):
	def _match_fft_norm(fft_norm_1, fft_norm_2, division=False):

then there would be no need to change the behaviour here.

[f-brinkmann]

@ahms5 implemented your changes as suggested. Now I'm waiting for the weekly before changing any tests