Decouple PATO from BFO and wire upper level to COB

[matentzn]

PATO's upper level is moving from BFO to COB (see #602). Cleanly cutting the BFO tie on PATO:characteristic (PATO:0000001) is blocked by an axiom outside our control: RO declares the range of has_characteristic (RO:0000053) as BFO:0000020. Until that range is changed to PATO:0000001, anything that imports RO will keep dragging BFO:0000020 back in and reasoners will keep classifying PATO:characteristic underneath it. Splitting this out of #602 so we can coordinate the cross-ontology change with RO and downstream consumers without holding up the rest of the upper-level reorg.

Context

The plan in #602 renames quality to characteristic and rewires the PATO upper level so the root no longer needs BFO:0000020 (specifically dependent continuant) as a parent. In principle, pato-edit.obo already does not assert PATO:0000001 is_a BFO:0000020 directly. In practice, reasoning over the release artifact still puts PATO:0000001 under BFO:0000020, for two independent reasons:

1. Foreign axiom in GO base. go-base.owl (versionIRI 2024-06-17) carries SubClassOf(PATO_0000001 BFO_0000020). This needs to be fixed upstream in GO: Remove BFO imports and rely on COB at the upper ontology geneontology/go-ontology#31354 (cc @balhoff)

2. RO range on has_characteristic. RO:0000053 (has characteristic) declares its range as BFO:0000020. As long as that holds, every ontology that uses has_characteristic (which is most of OBO) needs BFO:0000020 in scope, and PATO:characteristic will continue to be classified under it transitively. The fix is for RO to change the range to PATO:0000001 (PATO:characteristic).

Both fixes are cross-ontology and will break downstream tooling that assumes BFO:0000020 is the canonical type for characteristics/qualities. That is the reason for decoupling this from #602: #602 should land on its own timeline, and the BFO drop should be coordinated separately with RO, GO, and downstream consumers so they have notice.