Converted no-unnecessary-generics from TSLint to ESLint

[JoshuaKGoldberg]

Continues the work from #489.

Deduplicates the typescript package dependency versions by switching them all to the stable 4.7.4 (which was already being used). Without deduped versions, there are type errors like:

Argument of type 'Declaration' is not assignable to parameter of type 'Node'.
  Types of property 'kind' are incompatible.
    Type 'import("/Users/josh/repos/DefinitelyTyped-tools/packages/dtslint/node_modules/@typescript-eslint/typescript-estree/node_modules/typescript/lib/typescript").SyntaxKind' is not assignable to type 'import("/Users/josh/repos/DefinitelyTyped-tools/packages/dtslint/node_modules/typescript/lib/typescript").SyntaxKind'.
      Property 'OutKeyword' is missing in type 'import("/Users/josh/repos/DefinitelyTyped-tools/packages/dtslint/node_modules/typescript/lib/typescript").SyntaxKind'.ts(2345)

Also adds a tsconfig.json to the dtslint tests directory so that ESLint rules using type information can be tested. https://typescript-eslint.io/docs/development/custom-rules#testing-typed-rules

[JoshuaKGoldberg]

Bump @sandersn - is there any chance this & the other open TSLint->ESLint PRs be looked at soon, please?

[sandersn]

does this require a real 'file.ts' to exist? A quick read of the documentation suggests 'yes'.

Edit: But...all tests pass, so maybe not.

[sandersn]

missing the last example function example<T, U extends T>(t: T, u: U): U

[sandersn]

Suggested change

	function recurse(node: ts.Node): void {
	function recur(node: ts.Node): void {

The original verb form is 'recur'. But you can't have the noun form be *recurion, because, roughly, it violates a constraint on the pronunciation of the -ion suffix. 's' turns out to be the cheapest insertion that satisfies the constraint, giving you 'recursion'.

However, the noun 'recursion' gets used a lot more than the verb 'recur', so people end up treating that form as primary and creating the verb 'recurse' from 'recursion'. Chomskyan linguistics doesn't have a good explanation for this last part since it's not concerned with statistics or language change.

[JoshuaKGoldberg]

Haha, as much as I love discussion Chomskyan linguistics in PRs, this was the established convention in TSLint back in the day.

[sandersn]

Whew! Finally got a clean local run -- looots of exemptions on this one, in varying syntaxes. I'm going merge this now.