Python fsl.findimage() function

[Lestropie]

Another script fix necessary to get through my test of #1077.

[Lestropie]

I've successfully executed all scripts affected by these changes.

I also made a bad boo-boo in #1047 here: Because applytopup is now run separately for each phase-encoding direction, the list of its outputs may only contain one entry, and therefore mrcat can't be run... So this very much needs to go with 3.0_RC2: -rpe_none and -rpe_pair are currently broken. Have I mentioned that I resent dwipreproc? Perhaps it's time to make a matrix of tests (not necessarily in the CI) to make sure dwipreproc handles each possible case as it should; the different possibilities are so intertwined that it's really easy to fix one use case but break something else.

Open to review... Any volunteers?

[jdtournier]

Not sure I'm the best person to be reviewing this... I'm happy for it to be merged if you're 'happy' with it. As you say, this is the kind to things that requires extensive community testing to really identify all the edge cases. It would be good to have tests for all of these, but that will be a lot of work, and definitely won't make it into tag_3.0_RC2... So I'd merge now and add tests later if/when you get the chance...

[thijsdhollander]

Looks good, but agree with @jdtournier that we'll need to get "community debugging" for these things. Hence, I'd encourage the merge even more. And of course to fix -rpe_none and -rpe_pair... we can't have those being broken; they're the most frequently used scenarios, I reckon.