cmake tools性能问题

Type: Bug

vscode更新到1.102.3版本，cmake tools为1.21.36版本
打开一个以前已经构建过的cpp项目，在状态栏电点击启动所选目标，执行程序时，需要等待比以前更久的时间，才能完成构建和执行。
其中尤其卡在终端执行这一步，构建完成后会在终端运行.exe执行程序，但需要等很久才能看到输出。以前基本秒出程序运行结果

Extension version: 1.21.36
VS Code version: Code 1.102.3 (488a1f239235055e34e673291fb8d8c810886f81, 2025-07-29T03:00:23.339Z)
OS version: Windows_NT x64 10.0.26100
Modes:

System Info

|Item|Value|
|---|---|
|CPUs|Intel(R) Core(TM) i7-14700HX (28 x 2304)|
|GPU Status|2d_canvas: enabled
canvas_oop_rasterization: enabled_on
direct_rendering_display_compositor: disabled_off_ok
gpu_compositing: enabled
multiple_raster_threads: enabled_on
opengl: enabled_on
rasterization: enabled
raw_draw: disabled_off_ok
skia_graphite: disabled_off
video_decode: enabled
video_encode: enabled
vulkan: disabled_off
webgl: enabled
webgl2: enabled
webgpu: enabled
webnn: disabled_off|
|Load (avg)|undefined|
|Memory (System)|15.78GB (6.29GB free)|
|Process Argv|--crash-reporter-id 606a77bd-b962-48fd-9839-1a75c0ce5a57|
|Screen Reader|no|
|VM|50%|

A/B Experiments

```
vsliv368:30146709
pythonvspyt551cf:31249601
binariesv615:30325510
2e7ec940:31000449
nativeloc1:31344060
dwcopilot:31170013
6074i472:31201624
dwoutputs:31242946
copilot_t_ci:31333650
e5gg6876:31282496
pythoneinst12:31285622
c7cif404:31314491
996jf627:31283433
pythonrdcb7:31342333
usemplatestapi:31297334
0aa6g176:31307128
747dc170:31275177
aj953862:31281341
generatesymbolt:31295002
convertfstringf:31295003
pylancequickfixf:31358881
9d2cg352:31346308
convertlamdat:31358880
usemarketplace:31343026
nesew2to5:31336538
agentclaude:31352135
nes-diff-11:31337487
replacestringexc:31350595
testaa123cf:31335227
6abeh943:31336334
yijiwantestdri0626-c:31336931
0927b901:31350571
f76d9909:31348711
45650338:31358607
0cj2b977:31352657
05h3e699-none:31352670
gaj49834:31355361

```

Start by reproducing the delay on Windows with VS Code 1.102.3 and CMake Tools 1.21.36, using an already-built C++ project and the status-bar launch target. Compare the time spent building and waiting for terminal output with earlier versions; done means identifying a reproducible regression or narrowing the delay to a specific step.

Written by the indexing model from the issue text.