Remove the unconditional delay between Wave end-to-end tests

## Problem

`tools/run_tests.py` sleeps for 0.3 seconds after every test case. With more than 100 cases this adds over 30 seconds to every local and CI E2E run even when tests do not use sockets or shared resources.

This replaces the unbounded performance request in #118 with a measurable task.

## Scope

- Remove the unconditional per-test delay.
- If a platform-specific resource needs settling time, apply a targeted wait only to that resource and document why.
- Confirm that TCP/UDP tests continue to clean up their processes and sockets.
- Compare the E2E runner wall time before and after the change.

## Completion criteria

- [ ] Ordinary tests have no artificial inter-test sleep.
- [ ] Network tests remain stable on supported CI hosts.
- [ ] The complete test suite reports the same pass/skip/fail results.
- [ ] The PR reports before/after timing from the same host.

Start in tools/run_tests.py and measure the E2E runner wall time on the same host before changing the inter-test delay. Check the TCP/UDP tests for process and socket cleanup, then run the complete suite to confirm unchanged pass, skip, and fail results. Done means ordinary tests have no artificial sleep, network tests remain stable, and the PR reports before/after timing.

Written by the indexing model from the issue text.