ENH: Tiled QR wrappers for scipy.linalg.lapack

[AidanGG]

As per #10323 I have begun working on the wrappers. I believe the wrappers and documentation are complete.

I still need to write some tests (but they do appear to function according to my brief, informal testing).

Closes #10323

[AidanGG]

@ilayn I believe this is ready to be looked at, but there are some CI issues with the documentation? How would I fix this?

[ilayn]

Great. Thanks for the time and effort. I will check further today more carefully.

One of the travis failures is real though for float32 dtype. The fft stuff you can ignore for now; it is a new feature and CIs didn't refreshed apparently

[AidanGG]

Thanks, my test seems to fail on Travis with Python 3.5. I will investigate this locally with a new virtualenv.

In the mean time, I've pushed a change that adds a depends() to the f2py wrapper that might fix it.

Edit: This appears to have fixed it.

[ilayn]

Other than the minor issue, I don't have any other issues.

[ilayn]

The CircleCI stuff is irrelevant but in Travis run I think you are hitting the same issue that I had in #8965

The documentation is getting choked on the first unmatched * character of the size while trying to write the parameter bounds. You can use side[0] instead to escape from it.

[AidanGG]

Alright, these changes appear to have fixed the docs building.

[ilayn]

Thanks! Now it looks good to me, let's wait for a while in case others want to chime in.

[ilayn]

OK let's push this in. Since this is for 1.4 we would still have time to fix it in case I missed a detail. Thanks again @AidanGG, Keep them coming!

And then the obligatory copy/paste response:

Just for the future reference, please make separate branches to work on new features and to send PRs such that your standard working repo does not interfere with the PRs you have submitted. Once a PR is merged you can safely delete that branch and keep working on other branches.

[AidanGG]

Thanks, @ilayn. I'm thinking about adding some of the routines necessary for tiled Cholesky, which I'll get around to doing once I need them for my own work.