Use normal index deletion for child nodes

[CodingFeng101]

What changed

dom_tag.delete_attribute() now deletes child nodes with normal list index semantics when the key is an integer.

Why

Child reads and assignments already use normal list indexing: negative indexes work, and out-of-range indexes raise. Deletion was the odd path because it used a slice (key:key+1), which made del tag[10] silently do nothing and made del tag[-1] fail to remove the last child.

This keeps attribute deletion unchanged and only changes the integer child-index path.

Validation

• Reproduced locally before the change: del div("a", "b")[10] did not raise, and delete_attribute(-1) did not remove the last child.
• python -m pytest tests\test_dom_tag.py::test_delete_attribute_child_index tests\test_dom_tag.py::test_set_attribute tests\test_html.py::test_add -q
• python -m pytest tests -q
• python -m py_compile dominate\dom_tag.py tests\test_dom_tag.py
• git diff --check

Note: the repository-wide python -m pytest -q also collects README doctests locally; that currently fails on README doctest context setup (d is not defined), while the package test suite under tests passes.