[gui] Some fixes to TGNumberEntry

[silverweed]

An attempt to fix and improve some old code in TGNumberEntry by:

• fixing some broken logic
• simplifying some functions
• improving safety by adding bounds checks where appropriate

This is not an exhaustive pass on the file, as there are many other places that could use a pass or a rewrite. However it should address some of the more glaring safety concerns - particularly, it removes all uses of strcpy and the coverity annotations.

I haven't tested this code yet so I'm opening this as a draft PR. Feel free to comment on it already.

Checklist:

• tested changes locally
• updated the docs (if necessary)

This PR fixes #17334

[ferdymercury]

Thanks a lot for this initiative. If you want a reproducer to test against: https://igit.ific.uv.es/ferhue/pulse-surfer

[github-actions]

Test Results

    19 files      19 suites   3d 22h 21m 45s ⏱️
 2 731 tests  2 731 ✅ 0 💤 0 ❌
50 411 runs  50 411 ✅ 0 💤 0 ❌

Results for commit fc16914.

♻️ This comment has been updated with latest results.

[silverweed]

I tested on @ferdymercury 's reproducer and it seems to work fine.

[ferdymercury]

Looks great, thanks! Maybe just add a small doxygen documentation on what CopyAndEliminate is doing.

[dpiparo]

thanks for this. LGTM.

[ferdymercury]

@silverweed I am seeing a regression in TGNumberEntry, that might be connected with this PR.

In 6.36.00
root --web=off $ROOTSYS/tutorials/geom/building.C

then click on Clipping, Plane, numbers look fine.

In master:
root --web=off $ROOTSYS/tutorials/visualisation/geom/building.C

both in first tab and in Clipping, Plane, numbers look weird.
(I am rebuilding to see if sth went wrong with rebasing or so).