Signal1D/Signal2D merge into master

[dnjohnstone]

Following #963 to 0.8.x this is the merge in to master.... I will finish it by the end of the week.

[dnjohnstone]

Ok - this has become a total mess....

As I understood it the plan was to merge this in to master before anything else to avoid repeated merge frustrations. The day that we decided that this PR was made and was incredibly close except for the single Travis failure that I couldn't track on my machine because it wasn't failing on my machine.

Unfortunately, for some reason, loads of things have now been merged into both 0.8.x and master before the main Signal1D/Signal2D split was made. You'll see that the diff now stands at about 230 files which is going to make this incredibly difficult to review and this problem will only get worse if people keep merging things in to the master. Don't forget that every time you merge one of those PR's every file touched in both 0.8.x and that PR has to be remerged and since that's nearly everything in 0.8.x this creates a huge amount of work and only increases the likelihood of errors.

At this point I think that it might actually be best to start again with a fresh 0.8.x --> master merge because people have changed such a huge amount since this PR was made.

[francisco-dlp]

Currently tests are not passing due to i) external deprecation warnings ii) internal deprecations warnings. i) could be addressed in a different PR but ii) should be addressed in this one. Happily it seems that the non reproducible error is now gone, so it was probably due to a numpy bug.

The internal deprecation warning seems easy to fix : "The method AxesManager.connect() has been deprecated and will be removed in HyperSpy 0.10. Please use AxesManager.events.indices_changed.connect() instead."

[to266]

I've fixed the warning in dnjohnstone#9 however it does not fix the behaviour - the swapped axes do not seem to be correctly connected to the navigation events, and I could not figure it out for now (will try agian tomorrow if noone fixes it by that time)...

[dnjohnstone]

@francisco-dlp this is now passing the tests thanks to @to266.

As regards external deprecation warnings - maybe I don't understand them, but it seems like most of them are not our fault? e.g. travis picks up inspect.getargspec which should become inspect.signature, however, we only used this in one place and I've now made that change so I don't see how it can be us.

[francisco-dlp]

According to @to266 (see post above) the warning issue is now fixed but something is still wrong with the connection. Let's wait to hear back from @to266 tomorrow.

[to266]

The last issue I had with this is PR'ed here dnjohnstone#10

Once that is merged (edit: it is now!), I suggest someone merges this ASAP. Any leftover bugs will be much easier to fix in separate PRs

[dnjohnstone]

Thanks @to266 for the final fix, I really should learn about events better -- if you're happy with this @francisco-dlp I'd also be happy if this got merged.... I've been using this for all my day to day stuff for the past couple of days and haven't hit any bugs that I haven't already fixed.

PS... myself and Tomas did go through plotting tests as well.