Fix for CJSImportProcessor not using interop helper for aliased default imports

[Rugvip]

Hi! Found something that I think is a bug in the handling of aliased default imports.

For example, given the following code:

import { default as myFunc } from './myFunc';

myFunc();

We want this to be transpiled to something like this (simplified):

const myFunc = _interopRequireDefault(require('./myFunc'));

myFunc.default();

But since it's not treated as a default import, we get this kind of thing instead:

const myFunc = require('./myFunc');

myFunc.default();

Which may cause trouble if the imported module is not an ESM module.

[codecov]

Codecov Report

Merging #619 (aa6296e) into main (805cd66) will increase coverage by 0.09%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #619      +/-   ##
==========================================
+ Coverage   82.03%   82.13%   +0.09%     
==========================================
  Files          56       56              
  Lines        5618     5859     +241     
  Branches     1320     1321       +1     
==========================================
+ Hits         4609     4812     +203     
- Misses        725      763      +38     
  Partials      284      284              

Impacted Files	Coverage Δ
src/CJSImportProcessor.ts	92.30% <100.00%> (+0.15%)	⬆️
src/util/formatTokens.ts	70.58% <0.00%> (-10.06%)	⬇️
src/parser/util/identifier.ts	90.90% <0.00%> (-4.10%)	⬇️
src/util/shouldElideDefaultExport.ts	77.77% <0.00%> (-3.48%)	⬇️
src/util/getClassInfo.ts	87.21% <0.00%> (-1.93%)	⬇️
src/parser/plugins/jsx/index.ts	91.33% <0.00%> (-1.48%)	⬇️
src/parser/plugins/flow.ts	63.72% <0.00%> (-1.28%)	⬇️
src/identifyShadowedGlobals.ts	87.80% <0.00%> (-1.09%)	⬇️
src/parser/traverser/expression.ts	88.10% <0.00%> (-0.11%)	⬇️
src/Options.ts	100.00% <0.00%> (ø)
... and 20 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 805cd66...aa6296e. Read the comment docs.

[github-actions]

Benchmark results

Before this PR: 276.5 thousand lines per second
After this PR: 277 thousand lines per second

Measured change: 0.19% faster (1.64% slower to 1.33% faster)
Summary: Likely no significant difference

[alangpierce]

Good catch, and the fix looks great, thank you! I also checked the Babel and TS output and confirmed that both of them have this behavior as well.

[alangpierce]

Just published as 3.18.2