docs(tutorials): Tier-D reduction — 6 getting-started pages to 3

[Jaureguy760]

Summary

Follow-up to PR #99 (Tier-B). PI feedback was "still too much text" — this PR collapses the getting-started surface from 6 pages (848 lines) to 3 pages (~520 lines), a further 39% reduction on the tutorial section specifically.

Tutorial information architecture — before / after

Before (6 pages, 848 lines):

Page	Lines	Role
`quickstart.rst` (top-level)	66	"5-minute quickstart" — install + count + analyze
`tutorials/quickstart_counting.rst`	151	"Count in 5 minutes" worked example
`tutorials/bulk_workflow.rst`	175	Full bulk RNA + ATAC pipeline (shipped in PR #99)
`tutorials/scrna_seq.rst`	112	scRNA-seq workflow
`tutorials/scatac_workflow.rst`	156	scATAC-seq workflow
`tutorials/comparative_imbalance.rst`	188	Group comparisons

After (3 pages, ~520 lines):

Page	Lines	Role
`tutorials/bulk_workflow.rst`	175	Bulk RNA + ATAC pipeline (unchanged)
`tutorials/single_cell_workflow.rst`	165	NEW — scRNA + scATAC with data-type callouts
`tutorials/comparative_imbalance.rst`	188	Group comparisons (2 inbound refs updated)

Changes (3 commits)

1. c5dec76 — Delete top-level `quickstart.rst` (66 lines). Duplicated bulk_workflow intro; sole toctree entry removed.
2. 098496a — Delete `tutorials/quickstart_counting.rst` (151 lines). Redundant with bulk_workflow step 2.
3. 7f623fe — Create `tutorials/single_cell_workflow.rst` merging scrna_seq + scatac_workflow. Data-type difference reduces to GTF (scRNA) vs BED (scATAC) in `--feature`. Per-cell-vs-pseudobulk table preserved from old scatac; barcode-suffix troubleshooting preserved from old scrna. 7 inbound `:doc:` references updated across index.rst, choosing_workflow.rst, bulk_workflow.rst, faq.rst, user_guide/single_cell.rst, and comparative_imbalance.rst.

Verification

• `sphinx-build -q -b html` completes; Tier-D introduces zero new warnings (3 pre-existing duplicate-citation warnings for `Kumasaka2016` / `vandeGeijn2015` across methods/ pages shipped in PR docs: reduce text volume (~47%) + fix scientific correctness issues #99's Tier-B — already on origin/dev + master; out of scope here)
• `grep` for all deleted-page names in docs/ returns zero matches
• No Python changes → ruff/pytest/Rust tests are no-ops (but will run on CI)

Cumulative reduction (across Tier-B + Tier-D)

• Before Tier-B: ~6,300 lines across 30 files
• After Tier-B (PR docs: reduce text volume (~47%) + fix scientific correctness issues #99): ~3,360 lines across 25 files (−46.7%)
• After Tier-D (this PR): ~3,060 lines across 22 files (−51.4% cumulative)

Tutorial section alone: 3,495 → 520 lines (−85%).