Prettier munging

[gilch]

Closes #302.
Closes #289.
Closes #288.
Closes #278.

Deep and interrelated changes this time. Rewriting all the tests for each increment would have been a lot of extra work for little benefit, so I had to squash them. Rewriting them once was still a lot of work.

[gilch]

I'm kind of surprised that wasn't already covered. Looks like I need one more test for a couple of the new lines.

[gilch]

Hmm. OK. This is a bit more of an edge case than I thought when I saw the lines in isolation. It makes sense why it isn't covered now. Typically, we'd expect the compiler to have access to the same modules as run time, especially during incremental development. But this doesn't necessarily hold.

The compiler need not necessarily run on the same version of Python as what is used at run time, which could certainly imply a different set of installed modules. For example, the standalone property means you don't need the hissp package installed unless you're using it at run time.

But it could also go the other way, with the run time having access to extra modules. For example, modules might be dynamically generated as I'm considering in #280. The Brython example in the README would be another case. In that example, the compiler running on a standard Python distribution has no access to the browser. module, but it compiles Lissp code to Python that uses it on the front end. We can't have the compiler rewrite (browser..foo- to (browser.._macro_.foo- and then balk when it can't import the module, because it might be a normal function call that would work fine at run time. Compiling like a function call would be the right behavior.

I think the code is correct as-is and the try: statement with the uncovered lines can't simply be removed. I need to test this. I can probably just compile some code referring to a bogus module without actually running it, but make sure it looks like a function call. I'll try that.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 100.00%. Comparing base (15a3f8d) to head (2127fa1).

Additional details and impacted files

@@            Coverage Diff            @@
##            master      #307   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files            6         6           
  Lines          840       832    -8     
  Branches       113       113           
=========================================
- Hits           840       832    -8     

☔ View full report in Codecov by Harness.
📢 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.

[gilch]

That wasn't too bad. I'll squash that in too.