Optimize repo tree

[janikmu]

Hi maintainers,

while not necessarily blocking, while reviewing for JOSS (openjournals/joss-reviews#10439), the initial clone of the repository takes quite some time and the repo tree is huge (>300 MB):

janik@Mac Mighty % git-sizer --verbose
Processing blobs: 3054
Processing trees: 3136
Processing commits: 762
Matching commits to trees: 762
Processing annotated tags: 0
Processing references: 17
| Name                         | Value     | Level of concern               |
| ---------------------------- | --------- | ------------------------------ |
| Overall repository size      |           |                                |
| * Commits                    |           |                                |
|   * Count                    |   762     |                                |
|   * Total size               |   260 KiB |                                |
| * Trees                      |           |                                |
|   * Count                    |  3.14 k   |                                |
|   * Total size               |   926 KiB |                                |
|   * Total tree entries       |  24.5 k   |                                |
| * Blobs                      |           |                                |
|   * Count                    |  3.05 k   |                                |
|   * Total size               |   798 MiB |                                |
| * Annotated tags             |           |                                |
|   * Count                    |     0     |                                |
| * References                 |           |                                |
|   * Count                    |    17     |                                |
|     * Branches               |     1     |                                |
|     * Tags                   |     1     |                                |
|     * Remote-tracking refs   |    15     |                                |
|                              |           |                                |
| Biggest objects              |           |                                |
| * Commits                    |           |                                |
|   * Maximum size         [1] |  1.24 KiB |                                |
|   * Maximum parents      [2] |     2     |                                |
| * Trees                      |           |                                |
|   * Maximum entries      [3] |    37     |                                |
| * Blobs                      |           |                                |
|   * Maximum size         [4] |  47.2 MiB | ****                           |
|                              |           |                                |
| History structure            |           |                                |
| * Maximum history depth      |   597     |                                |
| * Maximum tag depth          |     0     |                                |
|                              |           |                                |
| Biggest checkouts            |           |                                |
| * Number of directories  [5] |   148     |                                |
| * Maximum path depth     [6] |     7     |                                |
| * Maximum path length    [6] |   114 B   | *                              |
| * Number of files        [5] |   419     |                                |
| * Total size of files    [7] |   592 MiB |                                |
| * Number of symlinks     [8] |     3     |                                |
| * Number of submodules       |     0     |                                |

[1]  a03e6b445822763735f66fc44079d53b186f1c6d
[2]  3511553f393b874254a1e73c0e23975547b31e9e (refs/heads/main)
[3]  71aa35353f494f39bb004dd7d9108dfcfd6757b9 (ed92c1afbb8e3476fdcdf71bf41dfeec0b3a12b3:main/.doctrees/api)
[4]  57e1839f2227d235ec5872592941f9716490e73b (refs/remotes/origin/experiments:hpo/SAC_SMAC/Walker2d-v4/seed_0/results.csv)
[5]  a335454a7f01d087396e3da1ee4cdf6f5cd578a4 (refs/remotes/origin/experiments^{tree})
[6]  c8db33de805963bc80763f3e5cbad3301f8da85c (refs/remotes/origin/joss_paper^{tree})
[7]  ff7087d26f8ebf336306882261b31dfed4e09474 (ca87d0f542b3a339a24d925c0abcdd38ceb7882c^{tree})
[8]  267a3c4e3108fc1c2855b5b1212ae8709c3b98b8 (refs/remotes/origin/experiments:hpo)

git-sizer shows that some experiment results have been committed to branches, maybe you can use something like https://github.com/newren/git-filter-repo to clean up a little. In my opinion this is relevant because mighty is not simply a module that is distributed over PyPi, but a framework where one needs to either clone this repository or use something like the template repo, which itself has some coldstart problems at the moment (automl/mighty_project_template#1).

I would think that if you promote the template repository to a real first-stop for new users such that they don't have to clone this repository the tree size becomes less of an issue 🙂 That would necessitate that the PyPI version is up-to-date (including things like the gymnasium versioning fix which is both in the joss_paper branch and in main but remains in the pip installable package!