Add support to generate locks and consume them from a different architecture

[648trindade]

Ticket #236

Adds the ability for conda-devenv to generate aarch64 or arm64 locks.
It also allows to generate locks for different architectures than the one from the host machine, and selects the correct lock when consuming it.

Fixes #236, #233. Closes #234.

[nicoddemus]

Thanks @648trindade for the PR!

Please take a look at my comments.

It also allows to generate locks for different architectures than the one from the host machine, and selects the correct lock when consuming it.

I don't see any changes in the PR regarding that, or am I missing something? Or no changes were needed, and you just wanted to emphasize this is possible?

[648trindade]

Thanks @648trindade for the PR!

Please take a look at my comments.

It also allows to generate locks for different architectures than the one from the host machine, and selects the correct lock when consuming it.

I don't see any changes in the PR regarding that, or am I missing something? Or no changes were needed, and you just wanted to emphasize this is possible?

The main change is in the rendering of the jinja dictionary on render_jinja method. conda-devenv was always querying processor-related information from platform.machine() instead of getting it from CondaPlatform enum that is passed. For instance, if you want to render a lock for a win-64 platform when running on a linux-aarch64 machine, it would select YAML entries with aarch64 instead of x86_64.

[nicoddemus]

Thanks @648trindade!

I will leave this open for a couple of days to give others a chance to review. 👍

[nicoddemus]

@648trindade @volpatto 3.5.0 released 🎉

conda-forge/conda-devenv-feedstock#43