fix(ai): build from the committed model catalog instead of refetching it (PrimeIntellect-ai#2038)

`npm run build` in packages/ai ran `generate-models` on every build, fetching
the live model catalog and rewriting `src/models.generated.ts` in place. When
an upstream source drops a model that the repo still references (currently
claude-sonnet-4.5, 28 references), every CI run fails `Build and check` and
`Test (ai)` regardless of the PR's content, and release builds
(build-binaries.yml) ship a catalog nobody reviewed.

Build now compiles the committed catalog only. Refreshing the catalog stays an
explicit, reviewed step: `npm run generate-models`. The generated file is the
script's only output and is committed, so clean checkouts build and test
without network access to the catalog sources.

Linear: RES-1269

Author: snimu

packages/ai/.changes/res-1269-deterministic-model-catalog-build.md

@@ -0,0 +1 @@
+- Changed the build to compile from the committed model catalog instead of refetching it; run `npm run generate-models` explicitly to refresh `models.generated.ts`.

packages/ai/package.json

@@ -65,7 +65,7 @@
 	"scripts": {
 		"clean": "shx rm -rf dist",
 		"generate-models": "npx tsx scripts/generate-models.ts",
-		"build": "npm run generate-models && tsgo -p tsconfig.build.json",
+		"build": "tsgo -p tsconfig.build.json",
 		"dev": "tsgo -p tsconfig.build.json --watch --preserveWatchOutput",
 		"dev:tsc": "tsgo -p tsconfig.build.json --watch --preserveWatchOutput",
 		"test": "vitest --run",