trees_gene_groups STEP_2: embed RUN_TAG in SLURM JobName for tier distinguishability

Both gene_groups-COPYME and gene_groups_hgnc-COPYME STEP_2 RUN-workflow.sh
hardcoded the SLURM JobName as

    s2_${tier}_blk_NN          (slurm-burst)
    s2_${tier}_${gg}           (slurm-standard)

where ${tier} is the internal small-vs-large split WITHIN one PARENT_RUN_NAME.
When a user runs the same STEP_2 workflow as multiple sibling RUN dirs
(e.g. workflow-RUN_1-tier_small_le1000 / workflow-RUN_2-tier_medium_le5000
/ workflow-RUN_3-tier_large_le10000, each pinned to large_threshold above
all entries so everything bins to "small"), all three dispatches produced
identical-looking JobNames (`s2_small_blk_NN`) — squeue couldn't visually
distinguish which RUN dir launched which job.

Manifested 2026-05-26: 57 + 97 + 41 = 195 hugo_hgnc STEP_2 jobs all queued
as `s2_small_blk_NN`, requiring scontrol show job + WorkDir inspection to
attribute each back to its source RUN. The 195 in-queue jobs were
renamed in-place via `scontrol update JobName=...` to confirm the new
scheme works visually (57 s2_RUN_1_small_blk_NN, 97 s2_RUN_2_small_blk_NN,
41 s2_RUN_3_small_blk_NN).

Fix: derive RUN_TAG from PARENT_RUN_NAME via

    RUN_TAG=$( echo "${PARENT_RUN_NAME}" \
               | sed -n 's|^workflow-\(RUN_[0-9]\+\).*|\1|p' )
    RUN_TAG="${RUN_TAG:-RUN}"

and embed it: `s2_${RUN_TAG}_${tier}_blk_NN`. Falls back to literal "RUN"
if the parent dir doesn't match the canonical workflow-RUN_<N>-* pattern
(no behavioral regression for non-standard names). Applied uniformly to
both slurm-burst (block_name) and slurm-standard (std_name + --output)
codepaths.

Same fix also propagated into the three in-flight hugo_hgnc RUN dirs
(gitignored research instance) so any re-dispatch picks up the corrected
naming without needing to re-cp from COPYME.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Author: 000specific