Add --arch global option to support arm64 macOS via Rosetta2

[errikos]

Right now, mx refuses to run on M1 Macs (#232).

Allow to specify a custom architecture target. Useful for arm64-based Apple computers, until all used native libraries are released for arm64.

Example usage:
mx --arch amd64 ...

[graalvmbot]

Hello Ergys Dona, thanks for contributing a PR to our project!

We use the Oracle Contributor Agreement to make the copyright of contributions clear. We don't have a record of you having signed this yet, based on your email address ergys -(dot)- dona -(at)- gmail -(dot)- com. You can sign it at that link.

If you think you've already signed it, please comment below and we'll check.

[errikos]

Hello Ergys Dona, thanks for contributing a PR to our project!

We use the Oracle Contributor Agreement to make the copyright of contributions clear. We don't have a record of you having signed this yet, based on your email address ergys -(dot)- dona -(at)- gmail -(dot)- com. You can sign it at that link.

If you think you've already signed it, please comment below and we'll check.

One of my organisations here at GitHub (HexHive at EPFL - my employer, https://github.com/HexHive) has signed the CLA.

[graalvmbot]

Hello Ergys Dona, thanks for contributing a PR to our project!

We use the Oracle Contributor Agreement to make the copyright of contributions clear. We don't have a record of you having signed this yet, based on your email address ergys -(dot)- dona -(at)- epfl -(dot)- ch. You can sign it at that link.

If you think you've already signed it, please comment below and we'll check.

[errikos]

As already stated above, one of my organisations here at GitHub (HexHive at EPFL - my employer, https://github.com/HexHive) has signed the CLA, so I have been told that it should not be necessary that I sign it individually.

[dougxc]

This change is too intrusive and risky to be made as an automatic feature. How about adding an --arch global option to mx to force the arch value selected?

[graalvmbot]

Ergys Dona has signed the Oracle Contributor Agreement (based on email address ergys -(dot)- dona -(at)- epfl -(dot)- ch) so can contribute to this repository.

[errikos]

• Added a global --arch option.
• Added a warning in case the specified architecture is different than the one detected by get_arch().
• Introduced a new function called get_effective_arch() to get the architecture that will be effectively used in operation. This is to keep the get_arch() semantics unmodified.
• Replaced the get_arch() calls with get_effective_arch().

There are currently no restrictions in the architecture that can be specified, except from the list of allowed options.

[dougxc]

We don't want to introduce a new concept of "effective arch" and now all callers now need to think about the diff between get_arch and get_effective_arch. Can you think of a scenario under which 2 different answers make sense in a single mx execution?

[errikos]

We don't want to introduce a new concept of "effective arch" and now all callers now need to think about the diff between get_arch and get_effective_arch. Can you think of a scenario under which 2 different answers make sense in a single mx execution?

To be honest the only reason for this change is to display the real underlying architecture in the introduced warning message.

We could instead just introduce an "internal" method _get_real_arch and leave the callers as they were.

[dougxc]

That looks much better.
@gilles-duboscq does this look good to you?

[gilles-duboscq]

LGTM

[dougxc]

force use the -> force use of the

[dougxc]

This PR now looks good to merge. Can you please rebase it on master and squash it to a single commit.