feat: Inject channels+mirrors coming from mambajs env-lockfiles

[Klaim]

Description

So far the handling of environment lockfile read package specifications from them and used that to get the packages but the channels urls specified in the lockfile were read but then ignored.

This change makes libmamba use the urls specified by lockfiles, only for mambajs for now, and put them at the top of the priority of urls to use for each channel name (could be easily changed later if necessary).

The core of the feature is located in MTransaction create_explicit_transaction_from_lockfile() (transaction.cpp), reviewers should start reading from there.

As part of that implementation:

• inject env-lockfile channels urls in Context and ContextChannel after reading the lockfile -- this unfortunately happens after the channels has been created with other sources of information so it is a bit contrieved by the execution flow;
• added contains standard algorithm implementation that will be replaced by the standard once available;
• replaced the channel selection resolution by a simpler shortcut based on the channel cache in ChannelContext (see the change in trancation.cpp in MTransaction::execute. This change has the potential to be breaking although we didn't find so far an explicit issue.
• changed the position of the publishing of json output in MTransaction::execute so that it relies on the deduced information instead of approximations pre-execution.
• ChannelContext::make_channel now allows adding new unique mirror urls to existing channels.
• Improved error information when failing to find package cache info.

Type of Change

• Bugfix
• Feature / enhancement
• CI / Documentation
• Maintenance

Checklist

• My code follows the general style and conventions of the codebase, ensuring consistency
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• My changes generate no new warnings
• I have run pre-commit run --all locally in the source folder and confirmed that there are no linter errors.
• I have added tests that prove my fix is effective or that my feature works
• New and existing tests pass locally with my changes

[codecov]

Codecov Report

❌ Patch coverage is 32.37410% with 94 lines in your changes missing coverage. Please review.
✅ Project coverage is 54.96%. Comparing base (b2d95cd) to head (3f533da).

Files with missing lines	Patch %	Lines
libmamba/src/core/package_cache.cpp	0.00%	28 Missing ⚠️
libmamba/src/core/transaction.cpp	10.71%	25 Missing ⚠️
libmamba/src/specs/channel.cpp	41.93%	18 Missing ⚠️
libmamba/include/mamba/specs/conda_url.hpp	0.00%	6 Missing ⚠️
libmamba/src/api/utils.cpp	0.00%	6 Missing ⚠️
libmamba/include/mamba/util/url.hpp	0.00%	4 Missing ⚠️
libmamba/src/api/install.cpp	0.00%	4 Missing ⚠️
libmamba/src/core/channel_context.cpp	83.33%	2 Missing ⚠️
libmamba/src/download/mirror_map.cpp	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4126      +/-   ##
==========================================
- Coverage   55.08%   54.96%   -0.12%     
==========================================
  Files         240      241       +1     
  Lines       30329    30424      +95     
  Branches     3241     3248       +7     
==========================================
+ Hits        16707    16723      +16     
- Misses      13619    13698      +79     
  Partials        3        3              

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[AntoinePrv]

I don't have the full context here, but if possible it would be best if these could be part of ChannelParams/resolve workflow.

The general idea of specs::Channel is that with resolve you always obtain what you want, (pure functional no mutable changes). This is because currently with libsolv hacks etc, specs::Channel are rather short-lived, converted back and forth to strings. I believe functions like set_platforms are use for only local reasoning.

[AntoinePrv]

If this logic could be made part of ChannelParams/Channel::resolve that would avoid adding extra complexity to the channel creation logic.

[Klaim]

I keep seeing issues with the tests locally but I suspect it's actually an unrelated problem, I might have to spam the ci a bit, sorry for the noise!

[jjerphan]

FYI, I also rebased this branch onto main.

I kept the previous state of the branch on backup/env-lockfile-channels.