[r/ci] Controlled downgrade for TileDB-R 0.23

[johnkerl]

Issue and/or context: Applies the same pattern as #1972 did for the release-1.5 branch and #1996 did for the release-1.6 branch.

Changes: As on #1972 and #1996. This is necessary since TileDB-R 0.24.0 now exists, and the R DESCRIPTION file does not syntactically support tiledb < 0.24 or tiledb <= 0.24. Therefore (see also this comment #1996 (comment)) we do a controlled downgrade to TileDB-R 0.23 on this PR, for the release-1.7 branch.

Notes for Reviewer:

Every stanza in this PR has been proved necessary via the dialog on #1972 and #1996. See in particular #1996 (review).

[codecov]

Codecov Report

Merging #2157 (24374c9) into release-1.7 (3c658a0) will decrease coverage by 6.55%.
The diff coverage is n/a.

Additional details and impacted files

@@               Coverage Diff               @@
##           release-1.7    #2157      +/-   ##
===============================================
- Coverage        78.53%   71.99%   -6.55%     
===============================================
  Files              136      101      -35     
  Lines            10804     7002    -3802     
  Branches           209      209              
===============================================
- Hits              8485     5041    -3444     
+ Misses            2219     1861     -358     
  Partials           100      100              

Flag	Coverage Δ
libtiledbsoma	67.59% <ø> (ø)
python	?
r	74.68% <ø> (ø)

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

Components	Coverage Δ
python_api	∅ <ø> (∅)
libtiledbsoma	50.47% <ø> (ø)

[eddelbuettel]

One difference is that because we do not have a CRAN release we also have no corresponding r2u or macOS binaries.
So would that mean that (for now) it is not (yet ?) a 'downgrade' ?

We may still want to impose the r-universe versions regardless.

[johnkerl]

@eddelbuettel This PR is not ready for review. Please do not review this PR.

I'll mark this PR as ready for review once its CI is complete, and once I've answered all questions I need to have CI answer for me.

[eddelbuettel]

@johnkerl I tried to explain how this current 0.24.0 is not like the previous ones.

[johnkerl]

@eddelbuettel thank you!

[johnkerl]

https://github.com/single-cell-data/TileDB-SOMA/actions/runs/7981104186/job/21792145630?pr=2157 shows that we are getting core 2.19.0 from TileDB-Py 0.26.0 but core 2.19.1 from TileDB-R 0.24.0:

Run python -c 'import tiledbsoma; tiledbsoma.show_package_versions()'
tiledbsoma.__version__        1.7.2.post4.dev2388502196
TileDB-Py tiledb.version()    (0, 25, 0)
TileDB core version           2.19.0
libtiledbsoma version()       libtiledb=2.19.1
python version                3.10.13.final.0
OS version                    Linux 6.2.0-1019-azure

tiledbsoma.__version__    1.7.2.post4.dev2388502196
tiledb.version()          0.25.0
core version              2.19.0
anndata.__version__  (ad) 0.10.5.post1
numpy.__version__    (np) 1.26.4
pandas.__version__   (pd) 2.2.0
pyarrow.__version__  (pa) 15.0.0
scanpy.__version__   (sc) 1.9.8
scipy.__version__    (sp) 1.12.0
python__version__         3.10.13

tiledbsoma:    1.7.2
tiledb-r:      0.23.0
tiledb core:   2.19.1
libtiledbsoma: 2.19.1
R:             R version 4.3.2 (2023-10-31)
OS:            Ubuntu 22.04.3 LTS

cc @ihnorton

[johnkerl]

@ihnorton :

• There are no problems here due to CRAN having TileDB-R 0.23.0 while https://tiledb-inc.r-universe.dev/tiledb has TileDB-R 0.24.0
• However, I believe we should have a TileDB-Py 0.25.1 with core 2.19.1 -- ?

[johnkerl]

Per @ihnorton there is no need for 2.19.1.

24374c9 ensures that we have 2.19.0 along all dependency paths.

[eddelbuettel]

I have a bit of a baked-in bias for 'newer/higher is better' but if it is felt that 2.19.0 everywhere is better / more feasible then this seems to do it so ...

[johnkerl]

@eddelbuettel same:)