fix: correctly parse unknown args with dashes when they resemble known args

[Xunnamius]

Resolves #501

Additionally, to compile yargs-parser and pass the CJS unit tests, I had to make two changes I did not include in this PR:

1. Upgrade to typescript@5 and add --noCheck to the "compile" and "pretest" scripts in package.json to work around some benign? unrelated type issues elsewhere in the codebase:

"scripts": {
...
-    "pretest": "rimraf build && tsc -p tsconfig.test.json && cross-env NODE_ENV=test npm run build:cjs",
+    "pretest": "rimraf build && tsc --noCheck -p tsconfig.test.json && cross-env NODE_ENV=test npm run build:cjs",
...
-    "compile": "tsc",
+    "compile": "tsc --noCheck",
...
  },
...
"devDependencies": {
...
-    "typescript": "^4.0.0"
+    "typescript": "^5.8.2"
  },

2. Remove transformDefaultExport-related configuration from rollup.config.js since the package seems to be non-functional; I don't believe the transform is required by non-EOL JS runtimes anyway:

import cleanup from 'rollup-plugin-cleanup'
import ts from 'rollup-plugin-ts'
- import transformDefaultExport from 'ts-transform-default-export'

const output = {
  format: 'cjs',
  file: './build/index.cjs',
  exports: 'default'
}

if (process.env.NODE_ENV === 'test') output.sourcemap = true

export default {
  input: './lib/index.ts',
  output,
  plugins: [
-    ts({
-      transformers: ({ program }) => ({
-        afterDeclarations: transformDefaultExport(program)
-      })
-    }),
+    ts(),
    cleanup({
      comments: 'none',
      extensions: ['*']
    })
  ]
}

[shadowspawn]

(Thanks for the notes on the hacks to run the tests. I did a draft of updates in #471 but have not looked at it since.)

[bcoe]

@shadowspawn I'll leave this PR to you, since it looks like it needs to be sequenced with your refactoring work?

[shadowspawn]

Oops, I forgot about this one. Thanks @Xunnamius , I will take another look.

[shadowspawn]

LGTM

The point of the regexp is a bit opaque to me, but based on the name of the variable and checking the behaviour in a regular expression tester, I think this make sense.

The new tests fail with the old regular expression and pass with the new regular expression.

[shadowspawn]

(I am having another look at isUnknownOption for other reasons. I'll see if I can make more sense of the purpose of the regular expressions and reconfirm the fix.)

[bcoe]

@shadowspawn perhaps we could add a performance test, like these, https://github.com/yargs/yargs-parser/blob/main/test/yargs-parser.mjs#L4005 that hits this regex specifically on a branch.

We don't need to fix the performance problem in this PR, but it's good to know if this PR incidentally addresses it.

[shadowspawn]

Undoing my approval as have a new PR candidate in #508

Have a newer PR to consider first.