[core] AddBefore/After/At option variant, as with Add/First/Last v2

[ferdymercury]

This Pull request:

Changes or fixes:

Fixes https://its.cern.ch/jira/browse/ROOT-9288

New version addressing this comment from pcanal: #17846 (comment)

Checklist:

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

[github-actions]

Test Results

    15 files      15 suites   3d 21h 58m 33s ⏱️
 2 688 tests  2 688 ✅ 0 💤 0 ❌
39 769 runs  39 769 ✅ 0 💤 0 ❌

Results for commit c3216c6.

[pcanal]

Rather than declaration this virtual we need to also update to TSeqCollection.h

[ferdymercury]

Thanks. I was just copy pasting the structure from lines 85-86. AddLast overrides from TSeqCollection, AddLast with option is virtual. Otherwise it's not consistent?

[ferdymercury]

And if we also moved AddLast and AddFirst to TSeqCollection to be consistent, that would break user-scripts deriving from that classes since they would become purely virtual (if one wants to keep consistent style).

[ferdymercury]

And if we made them non-purely-virtual, then users might start getting warnings such as warning: ‘virtual void ::AddAt(TObject*, Int_t, Option_t*)’ was hidden ?

[pcanal]

Otherwise it's not consistent?

Right. I missed that.

start getting warnings such as warning: ‘virtual void ::AddAt(TObject*, Int_t, Option_t*)’ was hidden ?

To extent that is better than use that:

• Inherits from TList
• Does not write/override AddAt(TObject*, Int_t, Option_t*
• Somehow starts using it ...
• It would 'silently' fail (depending what the overload does - but for example it would/was the case for THashList)

[ferdymercury]

v3 version here: #18389