test(one-home): dpkt's new _fragment helper's IP version checks are flagged as copies of IPV6_NEXT_HEADER_OFFSET #2342

tests/project/test_rfc_constants_one_home_1739_unit.py::TestNotReinlined::test_sweep fails on PR #2338's head (42903c784): the new pcapkit/toolkit/dpkt.py::_fragment helper takes an IP version number, and the sweep flags every 6 it spells as a copy of the homed constant IPV6_NEXT_HEADER_OFFSET (value 6, home pcapkit.protocols.internet.ipv6). A false positive by value collision with the unrelated IP version number 6.

Three sites, two distinct _context() strings:

pcapkit/toolkit/dpkt.py:708: _fragment(6) (a copy of a homed constant?)
pcapkit/toolkit/dpkt.py:569: version == 6 (a copy of a homed constant?)
pcapkit/toolkit/dpkt.py:561: version == 6 (a copy of a homed constant?)

The precedent is already in the test file's NOT_A_COPY dict for the sibling engine module:

('pcapkit/foundation/engines/dpkt.py', 'version == 6'):
    'the IP version number',

and the same reading for pcapkit/protocols/misc/pcap/frame.py and pcapkit/protocols/misc/pcapng.py, both ('…', 'version == 6'): 'the IP version number, not the decimal places of a resolution'. So pcapkit/toolkit/dpkt.py needs the equivalent two entries.

Failing job: Unit Tests / Python 3.15 (run 40417535198).

The fix cannot land on main ahead of #2338. test_sweep also guards against stale entries (assertEqual(sorted(set(NOT_A_COPY) - seen_copy), [])), so entries naming _fragment fail on a tree where _fragment does not exist yet. Measured on 9dd46d8d4: with the entries added and #2338's dpkt.py reverted, test_sweep fails with [('pcapkit/toolkit/dpkt.py', '_fragment(6)'), ('pcapkit/toolkit/dpkt.py', 'version == 6')] != []. The entries therefore have to go in #2338 itself, or merge simultaneously with it.