Resample to n_vols for sampling_rate == 'TR'

[adelavega]

WIP Fix to #712

• Keep track of n_vols in RunInfo (might be a good idea anyways)
• Use n_vols to compute the number of resample volumes in resample if sampling_rate == TR
• To do so, need to propagate the TR value one level up (instead of computing true sampling_rate at Collection level).

[adelavega]

We already have a test which catches the situation in which round does not return the expected index length.

I'm going to implement the special case scenario instead.

[effigies]

Failing test introduced in #365.

[effigies]

I don't think this is currently testing the edge case you brought up? Or can you comment on the lines that do?

Otherwise looks good.

[adelavega]

@effigies yep, still need to add a test. turned out to be a bit cumbersome but it seemed like adding n_vols to run_info was best, as it may end up being a good thing to have around anyways

[adelavega]

@effigies I added a test that sort of gets at it by setting a scan_length that is off by .1 (simulating the rounding error you sometimes see IRL). It's not the most robust test since it's not being read directly from the dataset. I think to do that however, we'd need to a new dataset with a TR below 1 and with some real images to read n_vol off. Not sure if worth it.

[adelavega]

This seems to be working for me. Care to review @tyarkoni ?

[tyarkoni]

LGTM!