test: enable 2 already-passing Node compat tests

[nathanwhitbot]

Summary

Enables 2 Node.js compat tests that already pass on main without any code changes. No production code touched — just tests/node_compat/config.jsonc entries.

• parallel/test-startup-empty-regexp-statics.mjs
• parallel/test-tls-handshake-exception.js

The .mjs regexp-statics test (unlike its .js sibling) skips the failing deno -p spawn assertion and only checks the static RegExp.$_ / RegExp.$1 / etc. defaults — which Deno gets right.

Test plan

• Each test verified passing 5/5 (mjs) and 3/3 (tls) times via the node_compat test runner
• Full cargo test --test node_compat run on Linux confirms no regressions (only pre-existing flaky failures)

[nathanwhitbot]

Force-pushed ebe93ba1d — branch was stacked on #33653 as a base. Reset to upstream/main and cherry-picked just the 2-test commit. Diff is now 1 file, +2 -0.

[nathanwhitbot]

Self-review of `ebe93ba1d` (post-cleanup):

Pure config-only test enrollment, no production code. Both entries alphabetically positioned correctly:

Test	Prev	Next
`test-startup-empty-regexp-statics.mjs`	`test-stack-size-limit.js`	`test-stdin-child-proc.js`
`test-tls-handshake-exception.js`	`test-tls-handshake-error.js`	`test-tls-honorcipherorder.js`

Both enabled with empty config (`{}`). Body's note about `.mjs` vs `.js` regexp-statics is accurate — the `.mjs` test only validates static `RegExp.$_`/`$1`/etc. defaults without spawning a `deno -p` subprocess that the `.js` version uses. Holding to COMMENT until CI confirms.

[nathanwhitbot]

Single-commit, 2-line config-only enrollment. No scope-creep. LGTM — TLS handshake test can be timing-sensitive on slower runners; if either test comes back red on Windows or macOS drop rather than gate.

[nathanwhitbot]

Both tests in this PR fail on windows-x86_64:

• test-startup-empty-regexp-statics.mjs — assertion failure (shard 1/3)
• test-tls-handshake-exception.js — timeout 10s (shard 3/3)

Drop both per @bartlomieju's convention.

(Shard 2/3 failure on test-dns-resolver-max-timeout.js is unrelated timing flake, not in this PR's scope.)

[nathanwhitbot]

Closing — both enabled tests fail on windows-x86_64 (test-startup-empty-regexp-statics.mjs assertion, test-tls-handshake-exception.js timeout). Per @bartlomieju's drop-rather-than-gate convention from #33593, dropping both leaves nothing to enroll. Will revisit individually if either turns out to be Deno-fixable rather than a genuine Windows incompatibility.