RawBinarySignalRawIO: Fix hard-coded dtype during memmap generation

[yagui]

fix #1764

[zm711]

Nice catch. Looks good to me. Doc error is unrelated. I'm running tests now, but we should be able to merge after that is done!

[zm711]

Also once we get this merged would you like to be on our author list? If so could I get the name you want to display and any institutional affiliation that should go with your name.

[h-mayorquin]

LGTM.

[h-mayorquin]

Maybe we should add to the todo to add a test that catches this and test other aspects of reading non-int16 binary data with this (say unsigned types) to see if we are not making any assumptions.

[zm711]

I think there is a test like that in spikeinterface where we have caught issues like this before, but since spikeinterface manages its own binary file reading we do a have blind spot for these rawios.

[samuelgarcia]

THanks.
Crazy that we catch this only now!!

[zm711]

@yagui we are about to do a new release. Just let us know if you want to be included in the author list for this release. (We can also add you later if you want :) )

[yagui]

Thanks, this was just one line. If I ever do a bigger contribution I'll be honored to be in the list. Thank you for this awesome library.