Fix operator precedence/associativity and support repeated percent postfixes

[ChasLui]

Motivation

• Correct operator reduction so groups have explicit associativity (notably making ^ right-associative) and ensure postfix % can be applied repeatedly.
• Simplify and centralize binary operator precedence logic to make maintenance and future changes easier.
• Add missing dev typing dependency for Node to align dev environment types.

Description

• Introduced OperatorAssociativity and BinaryOperatorGroup and replaced the previous precedence array with BINARY_OPERATOR_GROUPS that include associativity metadata.
• Replaced in-place precedence loop in formulaWithBinaryOp with reduceBinaryOperators and applyInfixAt helpers to correctly handle left- and right-associative operator groups (right-associative ^ reduced from the end).
• Changed percent handling from OPTION to MANY in formulaWithPercentOp so multiple % postfixes (e.g. 100%%) are parsed and applied sequentially.
• Added @types/node to devDependencies and updated pnpm-lock.yaml accordingly.
• Expanded operator precedence tests in test/operators/testcase.ts to cover chained exponentiation, mixed precedence/concatenation cases, various percent combinations, and chaining comparison behavior.

Testing

• Ran the unit test suite with vitest, including the test/operators cases; all tests passed.
• Confirmed the new operator precedence and percent chaining tests in test/operators/testcase.ts succeed under the updated parser implementation.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: dc7077722b

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Keep exponent chains left-associative

For formulas with two or more exponent operators, such as 2^3^2 or A1^B1^C1, marking ^ as right-associative makes the parser reduce the expression as 2^(3^2) and return 512. Excel evaluates operators of the same precedence from left to right, so caret chains should be reduced as (2^3)^2 and return 64; this breaks Excel-compatible results and the new tests lock in the wrong behavior.

Useful? React with 👍 / 👎.