Implement Test Runner (describe/it)

Implement solve(plan) where plan is a serialised test suite:

{
  suites: [
    {
      name: 'Math',
      beforeEach: 'counter++',   // JS expression appended to a harness
      afterEach: null,
      tests: [
        { name: 'adds', body: 'return counter === 1' },
        { name: 'still increments', body: 'return counter === 2' },
      ],
    }
  ]
}

Each beforeEach / afterEach / test body is a JS statement string
executed with new Function against a shared mutable context object
{ counter: 0 } that is reset to {} before each suite.

A test passes if its body returns true. Any other return value or a
thrown error means failure.

Return:

{
  passed: number,   // count of passing tests
  failed: number,   // count of failing tests
  results: [
    { suite: string, test: string, pass: boolean, error: string | null }
  ]
}

Rules:

• beforeEach runs before every test in the suite (including after failures).
• afterEach runs after every test, even if the test threw.
• If beforeEach throws, the test is marked failed with the error message; afterEach still runs.
• null means the hook is absent.