Revert "arch: change nxsched_suspend/resume_scheduler() called position"

[yamt]

Summary

This reverts commit 35c8c80.

because it caused a regression.
with esp32-devkitc:knsh, nsh prompt on serial console doesn't seem to accept any input.

Impact

Testing

tested with with esp32-devkitc:knsh

[nuttxpr]

[Experimental Bot, please feedback here]

The PR summary is extremely minimal. Here's a breakdown and suggestions for improvement:

What's good:

• It states the PR is a revert.
• It links the commit being reverted.
• It mentions the regression observed ("nsh prompt... doesn't accept input").
• It specifies the platform where testing occurred.

What needs improvement:

• Missing "Why": While stating it's a regression implies the change was detrimental, be more explicit. Was the original commit flawed? Did it break something unexpectedly? This adds context for reviewers.
• Vague Regression: "Doesn't accept input" is broad. Does nothing happen? Are characters echoed strangely? More detail helps pinpoint the issue.
• No "Before" State: Testing logs are empty. To prove a regression, show how the system behaved before the bad commit, then after. This demonstrates the change's impact.

Concisely, the PR does NOT meet NuttX requirements sufficiently. It's on the right track but lacks detail to make it easily reviewable.

[GUIDINGLI]

#13662

[GUIDINGLI]

@yamt
Can you try this PR:
#13662

[yamt]

@yamt Can you try this PR: #13662

i don't think i will test it anytime soon.
please test it by yourself or merge this revert. (i tested this)

[GUIDINGLI]

You can close this PR now, thanks

[xiaoxiang781216]

let's close this pr since the fix is merged.