feat: add support for computed values via skipped questions

[sisp]

I've added support for a long awaited feature: computed values via skipped questions (when: false). 🎉 See #1206 (comment) for more details.

Supersedes #1206. Resolves #1204 #1205. Resolves (I guess ...) #1203. 😜

[codecov]

Codecov Report

Merging #1220 (a491e48) into master (1b351bc) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1220   +/-   ##
=======================================
  Coverage   96.74%   96.75%           
=======================================
  Files          47       47           
  Lines        3935     3942    +7     
=======================================
+ Hits         3807     3814    +7     
  Misses        128      128           

Flag	Coverage Δ
unittests	96.75% <100.00%> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
copier/template.py	96.31% <ø> (-0.06%)	⬇️
copier/main.py	95.66% <100.00%> (+0.01%)	⬆️
copier/user_data.py	94.53% <100.00%> (-0.12%)	⬇️
tests/test_complex_questions.py	100.00% <100.00%> (ø)
tests/test_prompt.py	91.42% <100.00%> (+0.08%)	⬆️

... and 1 file with indirect coverage changes

[yajo]

Little improvement for docs. The rest looks great!

[yajo]

I'm thinking... We're just adding a feature, but what if a template relied on the fact that a skipped question was always missing? Should we label this as a breaking change? 🤔

[sisp]

Missing in the context? For this to be breaking, an injected variable through an extension like the context hooks one would need to have the same name as the skipped question. I guess, it can be breaking in rare cases. 🤔

I forgot to edit the inline comments to describe the new behavior. I'll push a commit with corrected comments in a moment.

[yajo]

Well, the feature was never supported, so after all any breakages would come from unsupported use. Thus we can consider this a normal improvement.

[sisp]

I think this PR isn't quite right yet. We should make sure that a skipped question's answer can only ever be its default value but not a value passed via --data or read from an edited answers file. I'll look into it.

[yajo]

not a value passed via --data or read from an edited answers file

I think you don't need to care so deeply for this. Those are either weird or unsupported use cases anyways. If you do them, you know you're gonna get something wrong, so it should be OK for us IMHO.

[sisp]

Alright, I guess we can keep it like it is now to ship it more quickly, possibly add more rigor later for those anyway weird cases.

[yajo]

It seems you have style violations. Could you run pre-commit please?

[sisp]

Ooops, yes, should be fixed now.

[pawamoy]

Best PR 🥲 Many users will be super happy about this!

[tlambert03]

thanks for your work on this @sisp!

I can confirm that with this PR, my template at https://github.com/pydev-guide/pyrepo-copier works as it did before v8

[sisp]

More important is @yajo's careful assessment about the best implementation of this feature.