`minimal` should be used only to set default feature

[pcanal]

In https://github.com/root-project/root/pull/20711/changes, in order to allow test to pass if (minimal) was used to turn off some of the test. However this is sub-optimal as:

• the test might actually work if the user turn back on the required features

and will lead to spurious failure if:

• the test still fails if the user disable the required features without using -Dminimal=ON.

[dpiparo]

Thanks for the 3 good remarks you made on the aforementioned PR. I did not see them before merging the PR. It's worth discussing them, even if I still believe that the PR as it was represented a good pragmatic solution for the item it addressed.

I would like to address each one of the comment made, proposing 3 pragmatic solutions:

• The TTabCom tests (here) failed for minimal builds because the class TH1Editor is in the /gui directory, which is added only for non minimal builds. To solve the issue, one can check at runtime if the TH1Editor class is available, and if not, adapt the expected tab-completion string.
• The optionlist test (here) is about a gui feature. However, as said above, there is no gui directory in minimal builds. Therefore, removing the test seems an adequate thing to do.
• The old formula1 tutorial (here) illustrates how a formula can be given in input to a TF1 function and drawn. I propose a simplification of the tutorial.

Before preparing a PR that then would be stuck, I would like to get a pre-approval of the ideas above, and discuss on the PR only implementation details, if any.