Fix broken stacks warning for universal traces

[Copilot]

When opening a universal trace in PerfView and viewing CPU stacks, PerfView incorrectly shows a warning about "100% broken stacks" even when the stacks are not actually broken. This happens because the broken stack detection logic was designed specifically for ETW traces but was being applied to all trace types.

Root Cause

The WarnAboutBrokenStacks method in PerfViewStackSource was running broken stack analysis on all trace types, but the logic for detecting broken stacks is only applicable to traces that originate from ETW.

Solution

Modified the WarnAboutBrokenStacks method to only perform broken stack analysis for ETW-based traces (ETLPerfViewData):

1. Added a type check to verify if DataFile is an instance of ETLPerfViewData
2. If not ETW-based, logs "Skipping broken stack analysis for non-ETW trace." and returns early
3. Preserves existing broken stack analysis logic for ETW traces where it's actually applicable

Impact

• Universal traces: No longer show false "100% broken stacks" warnings ✅
• EventPipe traces: Also benefit from not showing inappropriate warnings ✅
• Linux traces: Also benefit from not showing inappropriate warnings ✅
• ETW traces: Continue to show broken stack warnings when appropriate ✅
• Debugging: Added informative log message when analysis is skipped

Testing

Verified the type checking logic works correctly for different trace types:

• ETLPerfViewData (ETW traces) → runs broken stack analysis
• EventPipePerfViewData (universal traces) → skips broken stack analysis
• LinuxPerfViewData (Linux traces) → skips broken stack analysis

Fixes #2266.

---

💬 Share your feedback on Copilot coding agent for the chance to win a $200 gift card! Click here to start the survey.

[brianrob]

Copilot Post-Mortem:

• This one was pretty simple. A good example of the kinds of changes that copilot can knock out easily.