DOC: Clarify pitfall of ontological overreach in BEP definitions

[oesteban]

This PR adds a new entry under the “Common pitfalls” section of the BEP Guidelines to clarify the frequent issue of attempting to define neuroimaging concepts (e.g., “participant”, “atlas”, “template”) in ontological or universal terms.

While such definitions may be scientifically meaningful, they often delay progress and distract from the core goal of BIDS: to provide a practical and unambiguous data organization framework.

The added section recommends that BEPs focus on operational definitions—that is, defining entities and concepts only insofar as they are needed to support BIDS' internal consistency, validation, and tooling.

This clarification should help future BEPs avoid long-running definitional debates, particularly in areas like atlases, templates, provenance, and participant-related metadata.

No changes to schema or specification are included in this PR; it solely updates guidance in the BEP Guidelines.

[yarikoptic]

Overall I support promoting operational pragmatism and consistency! But, with

• #672

in mind, it gives me the neuroimaging-biased view toward future of BIDS development, which I would prefer to encourage.
In the long rung, if some inconsistency arises, I think we should rather strive to generalize definitions (e.g. as participants.tsv -> subject.tsv of bids-standard/bids-2-devel#14) or provide a modality specific explanation (which I believe we do where needed).

Hence whatever version is accepted here, IMHO it should be not neuroimaging-specific, and encourage succumbing to the definitions/meanings established in the early, neuroimaging-specific, days of BIDS. Rather I think it describe how specialized descriptions could be made, and describe on how/where breaking changes, if needed, should be introduced.