Drop python 3.10 and simplify for click 8.2 enum choice

[yarikoptic]

Summary

Follow-up to #1883 that collapses the click-8.2 compatibility shim
and rebases the CLI enums on the standard-library enum.StrEnum
(Python 3.11+). No behavior change — same command-line surface,
same --help output, same accept/reject semantics.

Two independent commits:

1. click 8.2 exposes Choice.normalize_choice as the explicit hook
   for choice-string normalization. Overriding it to return .value
   for Enum members (instead of the default .name) is all that
   is needed to preserve the pre-8.2 CLI surface, because
   Choice.convert then matches by normalized string and returns the
   matched enum member automatically. The custom __init__ +
   convert + self.enum_cls bookkeeping in Support click >= 8.2.0 #1883's EnumChoice
   become unnecessary. Bumps the click floor from >= 7.1 to
   >= 8.2 — the earlier bound predates normalize_choice.

2. Migrate the 11 CLI-facing (str, Enum) classes to
   enum.StrEnum and drop their hand-written
   __str__ -> self.value overrides. StrEnum was introduced in
   Python 3.11 and provides exactly these semantics natively. This
   requires bumping the minimum Python from 3.10 (EOL 2026-10-04)
   to 3.11: pyproject.toml, tox.ini, and 5 GitHub Actions
   workflows are updated accordingly (matrix drops the 3.10 leg;
   pinned single-Python include: entries move to 3.11).

Net change vs merged #1883: +39 / −83 across 14 files.

Commits

• b2d3c64 Simplify EnumChoice via click 8.2 normalize_choice hook
• 5d2cec5 Adopt enum.StrEnum for CLI choice enums; drop Python 3.10

Per-commit details

b2d3c64 Simplify EnumChoice via click 8.2 normalize_choice hook

EnumChoice shrinks from ~15 lines (custom __init__, convert,
self.enum_cls, defensive None/isinstance guards) to a single
3-line normalize_choice override. Callsites are unchanged —
iterating an Enum class yields members, and click.Choice.__init__
already does tuple(choices) on any iterable, so
EnumChoice(DownloadExisting) still works verbatim.

pyproject.toml: click >= 7.1 → click >= 8.2. The
normalize_choice hook is documented "added in 8.2.0" upstream;
retaining a lower bound would just be dead compatibility.

5d2cec5 Adopt enum.StrEnum for CLI choice enums; drop Python 3.10

Migrated: SyncMode (consts.py), DownloadExisting /
DownloadFormat / PathType (download.py), UploadExisting /
UploadValidation (upload.py), MoveExisting / MoveWorkOn
(move.py), FileOperationMode / CopyMode / OrganizeInvalid
(organize.py). Each gets its (str, Enum) base changed to
StrEnum and its __str__ -> self.value override deleted.

_Existing in dandi/cli/tests/test_base.py also migrated for
consistency.

EnumChoice itself is unchanged — StrEnum still exposes distinct
.name and .value, and click 8.2 still defaults to matching on
.name, so the normalize_choice override remains necessary.

Python floor moves from 3.10 to 3.11 to unlock enum.StrEnum:

• pyproject.toml: requires-python = ">=3.11", drop the 3.10
  classifier, [tool.hatch.envs.default] python = "3.11".
• tox.ini: drop the tomli; python_version < '3.11' conditional
  from the lint env (unreachable under the new floor).
• .github/workflows/run-tests.yml: matrix drops 3.10; the five
  include: entries pinned to '3.10' (dandi-api, dandi-api +
  ember-dandi, dev-deps, nfs, and the ember-dandi sandbox
  vendored config) move to '3.11'.
• .github/workflows/{lint,docs,release,typing}.yml: single
  python-version bumps to '3.11'.

Intentionally not migrated: dandi/utils.py:StrEnum (a
(str, Enum) shim used by dandi/validate/_types.py and
dandi/bids_validator_deno/_models.py). Its
_generate_next_value_ returns name (case-preserving), while
enum.StrEnum._generate_next_value_ returns name.lower().
Migrating it would change every auto()-derived value that appears
in serialized ValidationResult output (Standard.BIDS → "BIDS"
vs "bids", etc.) — a separate audit worth its own PR.

Test plan

• pytest dandi/cli/tests/{test_base,test_command,test_move,test_download}.py — all pass on click 8.4.2 + Python 3.13
• Broader dandi/cli/tests/ + dandi/tests/test_organize.py — 154 pass (only pre-existing bids-validator-deno-binary-dependent tests skipped in local env)
• dandi organize --files-mode SIMULATE still rejected; lowercase simulate still accepted
• dandi {download,upload,organize,move} --help render lowercase choice lists unchanged
• Pre-commit hooks (black, isort, codespell, flake8, reuse) pass on both commits
• Test locally interactively
• CI green on the PR (Ubuntu / macOS / Windows × Python 3.11–3.14)

[codecov]

Codecov Report

❌ Patch coverage is 47.61905% with 11 lines in your changes missing coverage. Please review.
✅ Project coverage is 76.93%. Comparing base (3f2b6d1) to head (b53ceba).

Files with missing lines	Patch %	Lines
dandi/organize.py	0.00%	5 Missing ⚠️
dandi/upload.py	0.00%	3 Missing ⚠️
dandi/consts.py	0.00%	2 Missing ⚠️
dandi/cli/base.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1890      +/-   ##
==========================================
+ Coverage   76.88%   76.93%   +0.04%     
==========================================
  Files          88       88              
  Lines       12901    12872      -29     
==========================================
- Hits         9919     9903      -16     
+ Misses       2982     2969      -13     

Flag	Coverage Δ
unittests	76.93% <47.61%> (+0.04%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 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.

[CodyCBakerPhD]

hint: if you tell your agent to 'watch the pr' it will automatically fix any CI failures, such as the docs build error

[yarikoptic]

thanks! I typically do not permit the beasts to push without my supervision, so would need another manual step in the loop

[github-actions]

🚀 PR was released in 0.76.7 🚀