docs: use "ID"/"IDs" consistently in documentation prose

[schochastics]

Fix #1683

Sweep roxygen prose to write "ID"/"IDs" instead of "id"/"ids" when referring to vertex and edge IDs, per the agreed convention. Lowercase is preserved where the token is code: argument names, runnable @examples, \preformatted and fenced code blocks, and `...` / \code{} spans.

Add tools/check-id-casing.R, a dependency-free checker that flags stray lowercase id/ids in roxygen prose, and document the convention in AGENTS.md and CONTRIBUTING.md.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 3632c85 is merged into main:

• ✔️as_adjacency_matrix: 755ms -> 757ms [-0.96%, +1.38%]
• ✔️as_biadjacency_matrix: 747ms -> 763ms [-0.66%, +4.82%]
• ✔️as_data_frame_both: 1.52ms -> 1.51ms [-2.44%, +1.2%]
• ✔️as_long_data_frame: 3.91ms -> 3.9ms [-0.8%, +0.32%]
• ✔️es_attr_filter: 2.75ms -> 2.77ms [-1.57%, +2.86%]
• ✔️graph_from_adjacency_matrix: 139ms -> 145ms [-4.43%, +13.86%]
• ✔️graph_from_data_frame: 3.42ms -> 3.42ms [-0.82%, +0.62%]
• ✔️vs_attr_filter: 1.63ms -> 1.67ms [-2.61%, +7.04%]
• ✔️vs_by_name: 1.01ms -> 1.01ms [-1.05%, +1.1%]
  Further explanation regarding interpretation and methodology can be found in the documentation.