Update codeblock answer substitution

[Bluesy1]

The changes here happened via cascade, I can split this into 4 separate PRs if wanted.

I recommend viewing the changes commit by commit, to understand what each is for:

• First I changed the codeblock substitution implementation, as per Support bacticks when answer options are created from a list, and other answer types #68
• Then, while testing this on a pl-matching question, I found and fixed an error with distractor answers (cc @amurph24 for https://github.com/open-resources/instructor_datascience_bank/pull/171#discussion_r1259040759)
• Then, testing that change, I noticed our tests weren't checking that question.html was being generated correctly, so I fixed that
• Which lead me to an error in my changes in Fix unicode support #51 , which is now fixed.

Resolves #68

[firasm]

These three can be handled in the same PR:

• fix to the unicode is fine, though I think we need to update our templates so there's a bit more complicated LaTeX in there so we don't keep missing weird edge cases (but this can be done in a later PR)
• fix to the matching question is also fine
• fix to question.html not being checked is also fine

---

• codeblock substitution is probably what we need to discuss
  • I don't like relying on pbs to be installed in courses, I'm actually trying to get to a point where only pbh is installed in courses as it's a much simpler dependency. Maybe we can move pbs.backticks_to_code_tags to pbh? (wherever pbs is installed, pbh` will likely exist, but not vice versa)

We should have a meeting about this to hammer out some things as it'll require some back and forth.

[Bluesy1]

• I don't like relying on pbs to be installed in courses, I'm actually trying to get to a point where only pbh is installed in courses as it's a much simpler dependency. Maybe we can move pbs.backticks_to_code_tags to pbh? (wherever pbs is installed, pbh` will likely exist, but not vice versa)

The alternative I had thought of was to just move the body of the convert function into the injected code, and inject an re import instead of a pbs import - I just wasn't sure if that would be too messy, injecting those 12 lines of code in. Moving it to pbh is a valid alternative.

If you want to have a meeting to discuss the different options, I can be available on Thursday for that.

I'll go ahead and split the changes that aren't for the codeblocks into a different PR so those can be merged in separately since this may take some time to figure out the optimal solution for

Edit: #70 has the split off changes

[Bluesy1]

I've rebased this off of #70 now, such that we can keep the changes in test running for anything made here, to be sure it works properly, but such that clean merging separately should hopefully be possible, though I suppose another rebase may be necessary down the line (should be relatively simple I think though)

[Bluesy1]

Change implemented as discussed earlier, this should not be merged (or at least merged and released) until a new release of PBH with open-resources/problem_bank_helpers#17 merged is cut (and ideally the dependency is bumped here, but that doesn't matter as much)