Remove fp_binary dependency from ddc driver

[JennySmith888]

fp_binary has not proved very useful and should ideally be removed from the requirements to enhance maintainability. Recommend replacing functionality with np.round, typecasting to np ints and using bit manipulation.

[baileyji]

I vaguely recall us using it in a few core cases early on to simulate fixed point math. Before we need to make a list of all code that uses it and decide what do do with it, maybe hiving that off into mkidanalysis or some other optional package. I think removing support from the drivers and any FRS paths would be relatively straightforward. That would open a third door so that it becomes an "optional" package that is only required for the python implementation of the fixed point opfb/fir. Please do not assign this to me, I'm not clear it is on a critical path.

[JennySmith888]

I understand your concern about it not being on the critical path. I was hoping you could help with this because you are the owner of the DDC and just refactored the driver and so the code is fresh in your mind. I understand why it is the way it is given the development history, but the code is not the cleanest and easiest to follow. For example: subclasses that don't seem to serve a purpose anymore, no type hints, apparent inconsistencies between allowed values of certain inputs and whether it is enforced with an assert, a doc string, etc. I can figure it out but it will probably take me longer and I'd like to not leave it in this state when I graduate.

[baileyji]

Got it.

Subclasses of what? Are you referring to the older DDC drivers? I think we could axe them and relegate them to either a tagged commit or moving the python to the hls repo. Without specifics examples any refactoring I do might not address what you see as a source of confusion.

I'll look at pulling it out of all the current code and leaving it as an optional needed for the software implementation of opfb. I think that shouldn't be too hard and is a reasonable goal that doesn't lose valuable history.

[JennySmith888]

Yes the older DDC drivers. I don't think the DDC driver needs to do any subclassing anymore. I think your proposal of tagged commit makes the most sense. That plus eliminating the fp_binary dependence is all the refactoring I am proposing.