ci: disable Bioconductor on older R so pak can resolve dependencies

[krlmlr]

LLM disclosure: this PR was prepared with Claude Code (Claude Opus 4.8).

• By submitting this pull request, I assign the copyright of my contribution to The igraph development team.

Problem

graph is an optional Enhances: dependency published only on Bioconductor. The Bioconductor releases matching R 4.2 / 4.3 (graph 3.16 / 3.18) are archived and now return HTTP 404 on the Posit mirror, so pak dependency resolution aborts on the rcc: … (4.2) / (4.3) jobs — they fail in the install step, before anything is built.

Fix

Rather than patching DESCRIPTION, flip pak's own switch: the install composite action sets PKG_USE_BIOCONDUCTOR=false when R < 4.4.

With Bioconductor off, graph has no install candidate; because it is a soft (Enhances) dependency, pak's solver drops it rather than erroring. Newer R keeps Bioconductor and exercises graph as before. DESCRIPTION is untouched.

Verified against pak's own source (the docs site 403s from here):

• pkgcache:::default_use_bioconductor reads the pkg.use_bioconductor option / PKG_USE_BIOCONDUCTOR env var.
• pkgdepends:::pkgplan_i_lp_dependencies explicitly nexts past a soft dependency whose candidates are all unavailable — so this drops graph silently rather than trading a 404 for a "package not found" error.

Also verified both branches of the version check locally (R < 4.4 writes the flag to GITHUB_ENV; R ≥ 4.4 writes nothing).

Evidence it works

In a full-matrix push run on an earlier revision of this branch, ubuntu-26.04 (4.2) / (4.3) no longer fail in install — dependency resolution now succeeds and they progress to the check step. (They still fail there, for a separate reason that is not dependency resolution; tracked separately.)

Scope

Workflow change only. The related pieces are split out:

• fix: use canonical structure() attribute names #2755 — the code changes (canonical structure() attribute names), off main.
• ci: run R-CMD-check on claude/** branches #2754 — run R-CMD-check on claude/** branches (stacked on this PR).
• fix: declare R_getRegisteredNamespace() so the C23 sanitizer build compiles #2749 — the R_getRegisteredNamespace C23 sanitizer error, which is a stale CSAN Docker image, not a package bug.

🤖 Generated with Claude Code

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 1d82032 is merged into main:

• ✔️as_adjacency_matrix: 685ms -> 683ms [-1.7%, +1.38%]
• ✔️as_biadjacency_matrix: 635ms -> 630ms [-2.43%, +0.75%]
• ✔️as_data_frame_both: 1.4ms -> 1.41ms [-1.11%, +2.39%]
• ✔️as_long_data_frame: 3.66ms -> 3.63ms [-2.3%, +0.82%]
• ✔️es_attr_filter: 2.65ms -> 2.66ms [-1.92%, +2.12%]
• ✔️graph_from_adjacency_matrix: 127ms -> 126ms [-5.39%, +3.65%]
• ✔️graph_from_data_frame: 3.14ms -> 3.1ms [-3.83%, +1.1%]
• ✔️vs_attr_filter: 1.4ms -> 1.39ms [-2.41%, +1.12%]
• ✔️vs_by_name: 910µs -> 909µs [-3.12%, +2.93%]
  Further explanation regarding interpretation and methodology can be found in the documentation.

[github-actions]

This is how benchmark results would change (along with a 95% confidence interval in relative change) if 6566387 is merged into main:

• ✔️as_adjacency_matrix: 884ms -> 883ms [-2.86%, +2.72%]
• ✔️as_biadjacency_matrix: 814ms -> 799ms [-4.51%, +0.81%]
• ✔️as_data_frame_both: 1.7ms -> 1.7ms [-2.42%, +2.66%]
• ✔️as_long_data_frame: 4.22ms -> 4.24ms [-2.55%, +3.59%]
• ✔️es_attr_filter: 2.83ms -> 2.83ms [-2.59%, +2.44%]
• ✔️graph_from_adjacency_matrix: 147ms -> 148ms [-0.9%, +1.48%]
• ✔️graph_from_data_frame: 3.67ms -> 3.62ms [-3.07%, +0.27%]
• ✔️vs_attr_filter: 1.63ms -> 1.61ms [-7.47%, +4.37%]
• ✔️vs_by_name: 1.04ms -> 1.06ms [-1.22%, +4.7%]
  Further explanation regarding interpretation and methodology can be found in the documentation.