Python: Fix DevUI streaming memory growth and add cross-platform regression coverage

[eavanvalkenburg]

Motivation and Context

Streaming responses in DevUI could cause browser memory usage to grow until the page became unresponsive or was killed. This change bounds the frontend memory footprint during long streaming responses and keeps that scenario covered with a portable regression test.

Description

• reduce per-token streaming state churn by persisting only minimal resumable state instead of full event history snapshots
• throttle assistant text updates and buffer debug text deltas so DevUI no longer rerenders the full chat/debug pipeline on every token
• add a browser-based DevUI memory regression test and make its browser process inspection portable across macOS, Linux, and Windows
• regenerate the shipped DevUI frontend bundle with the updated behavior

Contribution Checklist

• The code builds clean without any errors or warnings
• The PR follows the Contribution Guidelines
• All unit tests pass, and I have added new tests where possible
• Is this a breaking change? If yes, add "[BREAKING]" prefix to the title of the PR.

[moonbox3]

Python Test Coverage Report •

File	Stmts	Miss	Cover	Missing
TOTAL	27254	3202	88%

report-only-changed-files is enabled. No files were changed during this commit :)

Python Unit Test Overview

Tests	Skipped	Failures	Errors	Time
5486	20 💤	0 ❌	0 🔥	1m 23s ⏱️

[Copilot]

Pull request overview

This PR addresses DevUI browser memory growth during long streaming responses by reducing per-token state churn in the frontend and adding a browser-based regression test to catch future regressions across platforms.

Changes:

• Refactors streaming state persistence to store minimal resumable state (sequence/message IDs + accumulated text), with throttled saves.
• Throttles assistant message text rendering and buffers debug text deltas to reduce per-token rerenders.
• Adds a cross-platform, browser-driven memory regression test for streaming scenarios.

Reviewed changes

Copilot reviewed 5 out of 6 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
python/packages/devui/tests/devui/test_ui_memory_regression.py	New browser/OS RSS-based regression test to detect unbounded renderer memory growth during streaming.
python/packages/devui/frontend/src/services/streaming-state.ts	Removes event-history persistence; adds minimal state helpers for resumable streaming.
python/packages/devui/frontend/src/services/api.ts	Updates streaming/resume logic to record minimal state and throttle localStorage writes.
python/packages/devui/frontend/src/components/features/agent/agent-view.tsx	Throttles assistant text updates during streaming to avoid rerendering on every token.
python/packages/devui/frontend/src/App.tsx	Buffers debug text delta events and flushes periodically to reduce debug panel churn.

Comments suppressed due to low confidence (1)

python/packages/devui/frontend/src/services/streaming-state.ts:173

• markStreamingCompleted() is now unused (no call sites in the frontend), and the streaming flow clears state instead. Consider removing this export (or reusing it consistently) to avoid dead code and confusion about whether completed states should be persisted or deleted.

export function markStreamingCompleted(conversationId: string): void {
  try {
    const existing = readStreamingState(conversationId);
    if (existing) {
      existing.completed = true;
      existing.timestamp = Date.now();
      saveStreamingState(existing);
    }
  } catch (error) {
    console.error("Failed to mark streaming as completed:", error);
  }
}