Dimitri Papadopoulos Orfanos

Rebased AND added the license files under `subprojects`, even when they are duplicates of files under `scipy`. I cannot entirely remove `LICENSES_bundled.txt` because it contains licensing information for: * [LAPJVsp...

Will add the `LICENSES` top-level directory. How to name the license files in it? * pybind11 [`LICENSE`](https://github.com/pybind/pybind11/blob/master/LICENSE) (BSD-3-Clause) * NumPy [`LICENSE.txt`](https://github.com/numpy/numpy/blob/main/LICENSE.txt) (BSD-3-Clause unless we copied code under other licenses...) *...

For LAPJVsp, I have created a specific `lap_sparse` subdirectory to put the code and license file together. Happy to change the name or the method as explained in https://github.com/scipy/scipy/pull/12541#issuecomment-3200365888. Needs...

Done. I have copied the LICENSE files verbatim from the respective GitHub repositories, just renamed them as suggested. While grep'ing `Copyright` in the source code, I could see some files...

The end result will be very close to the initial `LICENSES_distributed.txt` file, just: 1. split into a singe file for each 3rd-party piece of code, 2. without information about the...

Ah right, I missed that individual license files will be machine-parseable, unlike `LICENSES_distributed.txt`. That covers point 1. And a `LICENSES/README` file would cover point 2.

@rgommers I think I've implemented what you have asked for. I am happy to change/normalise `LICENSES/LICENSE_*.txt` file names, modify the format of `LICENSES/README.txt`, or other similar changes. Note there may...

Also, while libuv is licensed under MIT ([`LICENSE`](https://github.com/libuv/libuv/blob/master/LICENSE)), it appears to contain some pieces of code under different licenses ([`LICENSE-extra`](https://github.com/libuv/libuv/blob/master/LICENSE-extra)).

We could, but some maintainers don't like them: ``` PT011 `pytest.raises(ValueError)` is too broad, set the `match` parameter or use a more specific exception PT012 `pytest.raises()` block should contain a...

I'd rather avoid adding the `match` argument in this PR, changes are extensive as far as I can remember. As for the second two, they imply transforming large `pytest.raises()`/`pytest.warns()` blocks...