Improve performance when using condition_dict

[knyghty]

Fixes #134

Uses a bit of caching to reduce the number of calls needed when using condition_dict.

For us, this reduces the number of queries in one view (for each step of that view) from nearly 300 to around 17. It's possible it could be improved further but I think this is a dramatic enough of an improvement to be useful on its own.

All the tests pass and I checked my wizards were all working locally. I went for a slightly more realistic test rather than a minimal test, but I can change this if desired.

A simpler solution is to cache the all property. It works quite well, but there are still some extra queries when doing this.

[claudep]

Thanks for the patch, it looks good to me, even if some added complexity might be a bug source.

I'd like at least one other approval before pushing it.

[codecov]

Codecov Report

❌ Patch coverage is 88.23529% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 93.62%. Comparing base (b31df3c) to head (8b17cdb).
⚠️ Report is 7 commits behind head on master.

Files with missing lines	Patch %	Lines
formtools/wizard/views.py	88.23%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #291      +/-   ##
==========================================
- Coverage   93.80%   93.62%   -0.19%     
==========================================
  Files          11       11              
  Lines         533      549      +16     
  Branches       86       90       +4     
==========================================
+ Hits          500      514      +14     
- Misses         21       22       +1     
- Partials       12       13       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[knyghty]

@claudep I did run into an issue testing this with real code. It looks like django-two-factor-auth pops from the form_list and this breaks things. I've added a fix for that here but I'm not sure if manipulating the form list like this is or should be supported. But this should at least avoid breaking existing code. I'm going to try running this in production if it works out okay in testing and let you know how it goes as the performance improvements here are quite necessary for us.

Let me know if you think it should be fixed in another way.

[knyghty]

The coverage issue above should now be fixed, didn't notice it before, sorry!

FWIW I've been running this in production for a couple of weeks now and seems to be doing well.

[claudep]

Thanks! Could we have a second approval here? Hopefully we are more than 2 people following the activity on this package 😜