fix(dart): Read Windows total physical memory via GlobalMemoryStatusEx instead of wmic.exe

[cf-rhett]

📜 Description

On Windows, the pure-Dart IO device-context enricher (PlatformMemory) read TotalPhysicalMemory by shelling out to wmic.exe (wmic ComputerSystem get TotalPhysicalMemory /VALUE), falling back to powershell.exe when WMIC was absent. This replaces both process spawns with a direct kernel call to GlobalMemoryStatusEx from kernel32.dll via dart:ffi, reading MEMORYSTATUSEX.ullTotalPhys.

The returned value keeps the previous semantics (bytes), so device.memorySize is unchanged. The useWindowsWmci / useWindowsPowerShell probing, the WMIC/PowerShell helpers, and the wmic.exe/powershell.exe existence checks are all removed. Adds ffi: ^2.0.0 as a dependency — already used by sentry_flutter.

💡 Motivation and Context

WMIC is deprecated and being removed from current Windows installs. Two problems with the current approach:

1. Endpoint-security noise. Every enriched event triggers a wmic.exe (or powershell.exe) child process under the host application. EDR/XDR tooling flags these spawns, and users ask why the app is invoking a deprecated Windows utility.
2. Cost. Spawning a process to read a single integer is far heavier than a syscall — the caching added in [Flutter] startTransaction slows down grpc Requests on windows #2760 reduced how often it happens, but the spawn is still there on the first enriched event.

GlobalMemoryStatusEx is the documented Win32 API for this value, needs no child process, and works regardless of whether WMIC is installed. The library the current code was copied from (system_info2 / onepub-dev/system_info) still uses WMIC on its master branch, so there's no upstream version to lean on here.

💚 How did you test it?

Covered by the existing test/event_processor/enricher/io_platform_memory_test.dart, which asserts a non-null, positive TotalPhysicalMemory on Windows and Linux and null elsewhere. dart analyze and dart format are clean on the change.

📝 Checklist

• I reviewed submitted code
• I added tests to verify changes (existing test covers the Windows path)
• No new PII added or SDK only sends newly added PII if sendDefaultPii is enabled
• I updated the docs if needed
• All tests passing
• Public API changes reviewed by another Mobile SDK team member or implemented according to the develop docs spec
• No breaking changes

Changelog Entry

Read Windows total physical memory via the native GlobalMemoryStatusEx API instead of spawning the deprecated wmic.exe (or PowerShell) process.

[codecov]

Codecov Report

❌ Patch coverage is 25.00000% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 88.06%. Comparing base (d527ac5) to head (8041659).
✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
...c/event_processor/enricher/io_platform_memory.dart	25.00%	9 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3894      +/-   ##
==========================================
+ Coverage   87.94%   88.06%   +0.11%     
==========================================
  Files         347      347              
  Lines       12929    12906      -23     
==========================================
- Hits        11371    11366       -5     
+ Misses       1558     1540      -18     

Flag	Coverage Δ
sentry	87.86% <25.00%> (+0.23%)	⬆️
sentry_dio	97.73% <ø> (ø)
sentry_drift	93.57% <ø> (ø)
sentry_file	65.29% <ø> (ø)
sentry_firebase_remote_config	100.00% <ø> (ø)
sentry_flutter	92.16% <ø> (-0.01%)	⬇️
sentry_grpc	99.09% <ø> (ø)
sentry_hive	77.48% <ø> (ø)
sentry_isar	74.37% <ø> (ø)
sentry_link	21.50% <ø> (ø)
sentry_logging	97.01% <ø> (ø)
sentry_sqflite	88.81% <ø> (ø)
sentry_supabase	97.27% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

[lucas-zimerman]

Looks good to me, CI passed with the changes in regard to windows, there are some failures on other platforms but probably unrelated to this PR.

@buenaflor Anything that you would like to point out? I checked locally, and the PR is working as intended, if there is nothing else to change from your part, LGTM!

[buenaflor]

LGTM if you tested it as well. CI failures are because of missing auth for E2E tests, but that's fine

[buenaflor]

There seems to be Analyze errors