Fix sequtils.delete bug with out of bounds indexes

[GULPF]

The proc already handled the case where last is out of bounds, so this is imo just a bugfix.

Fixes #12388

[GULPF]

Previously sending first > last resulted in a IndexError, which doesn't tell the user much about what went wrong. If sending first > last is supposed to work I can fix it instead.

[cooldome]

IMO, first greater than last should silently delete nothing. This how you define empty sequence with inclusive ranges.

[cooldome]

And first >= len should be IndexError

[GULPF]

And first >= len should be IndexError

I actually agree, but it would be inconsistent with how the last parameter is handled (out of bounds allowed) and how system.delete behaves. If first >= len results in IndexError then last >= len should also result in IndexError, which would be a breaking change.

[Araq]

There is some Python influence here too, "oobasfd"[:10000] == 'oobasfd', Python is usually silent about index out of bounds for slices.

[narimiran]

The code example in #12388 shows that also system.delete has a bug where it wrongly deletes the last element of the seq if you pass an index which is out of bounds.

Can you also patch that one?

---

Btw, system.del has another behaviour — it throws IndexError.
Shouldn't all three (sequtils.delete, system.delete, system.del) be consistent with each other when an out of bounds index is passed?

[Araq]

This improves the situation so I'm merging this now and later on we can make these APIs more consistent with each other.