Remove deprecated option -noall-load

[tkralphs]

Per the explanation here, this option no longer works in current versions of MacOS and even in older versions was being ignored with a warning.

[tkralphs]

@dpo I'm trying to fix the issue with the mums-seq recipe, as reported in #91 because it is causing testing failures in building CyLP here.

It seems the fix in this comment is the right one, per the explanation here. But there is a weird (unrelated) failure with uploading artifacts at the end of the tests.yml action. The actions/upload-artifact action complains that we are trying to attach multiple artifacts with the same name to a workflow, as per the error here. I can't see how this could be the case.

After some Googling, I found that actions/upload-artifact@v3 doesn't complain about duplicate artifact names (it just overwrites the previously uploaded artifact), so it's possible this was an issue all along and only manifested itself now that v4 is the default. I temporarily specified v3 and everything seems fine---I see three bottles with different names. This is weird, since it means the three bottles are produced that do indeed have different names.

I can certainly dig into this further myself, but thought I would see if you can make an educated guess, since you know the details of what's going on here a little better. I dug around a bit and couldn't find any relevant issues anywhere, which is weird. I did also recently update the platforms on which we're doing out CI to include newer MacOS. I noticed one anomaly---it's a bit strange that the bottle names are different on MacOS 14 versus MacOS 13:

mumps-seq--5.5.1_1.arm64_sonoma.bottle.tar.gz
mumps-seq--5.5.1_1.ventura.bottle.tar.gz

There is no architecture specified for the Ventura build. But this is a red herring I guess. Anyway, let me know if you have any ideas.

[tkralphs]

I tried running each platform separately and each of those runs worked and produced the expected file. None had the same names. I tried to disable the caching, since I thought maybe the issue is that the build outputs are being cached. But that didn't fix anything. I'm at a bit of a loss and tempted to just move on using v3.