feat(template): add axis-decomposed process_cpus_* / process_mem_* labels

[pinin4fjords]

Summary

Template change for nf-core/proposals#139 (approved Stage 2, 10/10) and the corresponding Stage 3 RFC at nf-core/proposals#140. Companion docs PR: nf-core/website#4212.

Introduces axis-decomposed resource labels alongside the existing bundled ones:

• process_cpus_{single,low,medium,high} - cpus only
• process_mem_{low,medium,high} - memory only
• process_time_{short,medium,long} - time only

A process stacks one of each axis to express its resource shape independently. e.g. a cpu-bound, memory-light, fast tool uses process_cpus_high + process_mem_low + process_time_short.

The bundled labels (process_single, process_low, process_medium, process_high, process_long, process_high_memory, process_low_memory) stay in place for backwards compatibility. Per the RFC, deprecation follows in a later PR once nf-core/modules and the institutional configs have migrated.

This supersedes the short-term process_low_memory workaround merged in #4264.

Motivation

Bundled labels couple cpus, memory and time at fixed ratios (e.g. process_high pins all three at 12 cpus / 72 GB / 16 h), which is the wrong shape for cpu-bound memory-light tools (Rust streaming binaries - trim_galore 2.x peaks at ~100 MB on 30M PE in ~1.5 min) or memory-hungry single-threaded ones. Splitting the axes lets module authors pick each independently and stops over-allocating on one axis just to get headroom on another. Full rationale and the agreed migration path live in the RFC.

Values

withLabel:process_cpus_single  { cpus   = { 1                    } }
withLabel:process_cpus_low     { cpus   = { 2     * task.attempt } }
withLabel:process_cpus_medium  { cpus   = { 6     * task.attempt } }
withLabel:process_cpus_high    { cpus   = { 12    * task.attempt } }
withLabel:process_mem_low      { memory = { 1.GB  * task.attempt } }
withLabel:process_mem_medium   { memory = { 12.GB * task.attempt } }
withLabel:process_mem_high     { memory = { 72.GB * task.attempt } }
withLabel:process_time_short   { time   = { 1.h   * task.attempt } }
withLabel:process_time_medium  { time   = { 8.h   * task.attempt } }
withLabel:process_time_long    { time   = { 20.h  * task.attempt } }

• CPU values are 1:1 with the existing bundled-label cpus tiers.
• Memory covers a wide range, with mem_low set sub-process_single (1 GB rather than 6 GB) so streaming tools have a meaningful low-memory bucket.
• Time uses 1.h / 8.h / 20.h. short = 1.h is a meaningful step down from the 4 h template default; medium = 8.h and long = 20.h match the existing process_medium and process_long values exactly so migrating tools don't change their time budget.
• process_high_memory (200 GB) is not folded in by this PR. It stays as a separate niche label until we decide whether a mem_xl tier is warranted (open question in the RFC).

Changes

• nf_core/pipeline-template/conf/base.config - new axis labels added above the bundled block, with a header comment marking them preferred and the bundled block marked as pending deprecation.
• nf_core/modules/lint/main_nf.py - extend correct_process_labels to accept both new and bundled labels.
• nf_core/components/create.py - extend the nf-core modules create autocomplete with the new labels.
• CHANGELOG.md - entries under Linting and Template.

Out of scope (tracked in the RFC)

• Deprecation lint warning and eventual removal of bundled labels.
• Three-question dialogue in nf-core modules create (currently still single-label autocomplete).
• Bulk migration in nf-core/modules and nf-core/configs.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 78.32%. Comparing base (4c78254) to head (44476ca).

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.