Remove explicit any usage and enforce stricter lint/type rules

[MontrealAI]

Motivation

• Eliminate widespread use of any casts and unsafe type assertions to improve runtime safety and enable stronger static checks.
• Introduce explicit unknown handling and typed guards so data parsed from external sources is validated before use.
• Harden lint and TypeScript configs to prevent regression by surfacing any usage and unsafe catch typings.
• Keep existing behavior unchanged while making the codebase safer and easier to reason about.

Description

• Replaced any and as any casts with unknown-aware code, typed guards (isRecord) and safe conversion helpers across orchestrator, validator, console, onebox, arena and shared modules (examples: apps/orchestrator/*, apps/validator/*, apps/console/*, services/arena/*, shared/*).
• Tightened runtime parsing and Zod schemas to use z.unknown() instead of z.any() for input records and payloads (files: packages/onebox-sdk/src/types.ts, packages/onebox-orchestrator/src/ics/schema.ts, packages/orchestrator/src/ics.ts, services/arena/src/validators.ts).
• Added small helper types and declarations: isRecord guards, toStringValue, typed harness interfaces for tests, and EIP-1193 window.ethereum typings (apps/console/src/types/ethereum.d.ts, apps/validator-ui/types/ethereum.d.ts).
• Enforced rules via config changes: tsconfig.json (noImplicitAny, useUnknownInCatchVariables) and eslint.config.js (set @typescript-eslint/no-explicit-any to error globally and permitted any in test/demo paths), and minor Solidity cleanup (removed an unused ERC20 import).

Testing

• Ran npm run lint:fix which executed ESLint/Prettier fixes and Solhint (warnings only); the lint fix command completed successfully.
• Ran TypeScript checks for webapps: npm run typecheck --prefix apps/console, --prefix apps/enterprise-portal, and --prefix apps/onebox and they completed without errors.
• Ran the test suite via npm test; unit and integration suites executed and the Hardhat tests completed (Hardhat run finished with tests passing).
• All gates requested (lint:fix, typecheck, and test) were exercised and completed successfully in the run.

---

Codex Task

In general, this kind of issue is fixed by either (a) removing redundant comparisons that are already guaranteed by previous checks, or (b) restructuring the condition to ensure that values of incompatible types are not being compared in a meaningless way. Here, the function already returns early when packed is null or undefined, so the later comparison packed !== null within the typeof packed === 'object' branch cannot ever be false. That makes the packed !== null part redundant and is what CodeQL is complaining about.

The best fix is to simplify the object branch condition from:

} else if (
  typeof packed === 'object' &&
  packed !== null &&
  'toString' in packed &&
  typeof packed.toString === 'function'
) {

to:

} else if (
  typeof packed === 'object' &&
  'toString' in packed &&
  typeof packed.toString === 'function'
) {

Because null has already been handled at the top of the function, this change does not alter runtime behavior but removes the unnecessary comparison between packed and null. No new methods, imports, or definitions are needed; this is a purely local condition simplification within apps/orchestrator/bidding.ts in the decodePackedJobMetadata function.