CLUE-406 chore: replace ts-loader with swc-loader for ~3x faster builds

[scytacki]

Summary

• Replace ts-loader with swc-loader (Rust-based transpiler) for ~3x faster dev builds
• Add ForkTsCheckerWebpackPlugin to run TypeScript type checking in a parallel process
• Configure ESLintPlugin with lintDirtyModulesOnly in dev mode
• Retain ts-loader for CODE_COVERAGE builds (Istanbul instrumentation requires it)

Build performance improvement:

Metric	Before	After	Improvement
Dev initial build	~31s	~10s	3x faster
Core file rebuild	~2.2s	~0.7s	3x faster

Test plan

• Verify npm start works and HMR rebuilds are fast
• Verify TypeScript errors still appear in the browser overlay
• Verify ESLint errors still appear in the browser overlay
• Verify CODE_COVERAGE=true npm start still works (uses ts-loader)
• Verify npm run build succeeds
• Verify npm test passes

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 66.75%. Comparing base (2aaec63) to head (2113d5e).

❗ There is a different number of reports uploaded between BASE (2aaec63) and HEAD (2113d5e). Click for more details.

HEAD has 24 uploads less than BASE

Flag	BASE (2aaec63)	HEAD (2113d5e)
jest	2	1
cypress-smoke	2	1
cypress-regression	15	0
cypress	7	0

Additional details and impacted files

@@             Coverage Diff             @@
##           master    #2756       +/-   ##
===========================================
- Coverage   86.81%   66.75%   -20.06%     
===========================================
  Files         829      824        -5     
  Lines       44549    44526       -23     
  Branches    11462    11460        -2     
===========================================
- Hits        38673    29723     -8950     
- Misses       5527    13767     +8240     
- Partials      349     1036      +687     

Flag	Coverage Δ
cypress	?
cypress-regression	?
cypress-smoke	43.89% <ø> (-0.27%)	⬇️
jest	49.85% <ø> (ø)

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

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[cypress]

collaborative-learning    Run #17660

Run Properties:   Passed #17660  •  2113d5e0e3: CLUE-406 chore: replace ts-loader with swc-loader for ~3x faster builds

Project	collaborative-learning
Branch Review	CLUE-406-swc-loader-faster-builds
Run status	Passed #17660
Run duration	02m 58s
Commit	2113d5e0e3: CLUE-406 chore: replace ts-loader with swc-loader for ~3x faster builds
Committer	Scott Cytacki
View all properties for this run ↗︎

---

Test results
Failures	0
Flaky	0
Pending	0
Skipped	0
Passing	4
View all changes introduced in this branch ↗︎

[kswenson]

Heads up before this merges: it reproduces a bug we just fixed in CODAP v3, in
concord-consortium/codap#2672.

The problem: swc-loader does not read tsconfig.json. With no jsc.target in the loader
options, swc falls back to its ES5 default — down-leveling const→var, expanding optional
chaining, nullish coalescing, arrow functions, class, generators, etc. CLUE's
tsconfig.json sets "target": "ES2015", so ts-loader emits ES2015 today; swapping in
swc-loader as written here silently drops the app bundle to ES5. That's transpiler bloat
downloaded by 100% of users, and it doesn't show up as a build error or a test failure — the
diff looks fine and the app still runs.

For scale: in CODAP, going from swc's ES5 default to es2020 cut the app chunk by
674 KB raw / 98 KB gzip (−8.3% / −4.5% of total JS) with no change in build time. CLUE's
src is comparable in size, so expect the same order of magnitude. Only the app chunks are
affected — swc only processes src, so vendor chunks are unchanged.

Suggested fix: set jsc.target, and derive it from tsconfig.json rather than hardcoding a
second value that's free to drift the next time tsconfig.json changes. What CODAP does
(see the diff in that PR):

• read the target through the TypeScript compiler API — ts.readConfigFile(...) rather than
  require(), because tsconfig.json has comments in it and require() would throw;
• throw on a missing or unrecognized target instead of falling back, since a silent
  fallback lands right back on ES5 — the exact bug being fixed, and one that is invisible in a
  diff.

A couple of CLUE-specific notes:

• The derived value here would be es2015, which just restores parity with what ts-loader
  emits today. Whether to raise tsconfig.json to es2020 is a separate call — worth making,
  but it changes tsc/ts-jest output too, so probably its own story. (There's a
  browserslist in package.json — "> 0.2%, last 5 versions, Firefox ESR, not dead" — but
  nothing wires it to swc, and not dead is a much lower floor than ES2020.)
• CODAP also had to add tsconfig.json to cache.buildDependencies so that changing the target
  invalidates webpack's persistent cache. CLUE has no filesystem cache in webpack.config.js
  today, so that's not needed now — just something to remember if one gets added.
• The CODE_COVERAGE branch keeps ts-loader, and Jest goes through ts-jest, which reads
  tsconfig.json. Both are unaffected either way.

How to verify: run a production build and grep the app chunk for ??, ?., =>, class,
and const. If they've all been rewritten into helpers and var, the target isn't taking
effect.