feat: allow to emit activation environment variables as JSON

[borchero]

Motivation

Fixes #1298.

Changes

After talking to @pavelzw, I decided to do the following:

• Introduce the option to emit env vars as JSON as additional flag under the shell-hook command, namely pixi shell-hook --json (in which case the choice of --shell doesn't matter)

• Do not handle PATH in a special manner as third-party tools can still do so if required. This makes the output format maximally flexible and easily allows for augmenting it e.g. in the context of docker support: Let pixi create images #1118. The output format is now simply:

  {
    "environment_variables": {
      "<key>": "<value>"
    }
  }

[ruben-arts]

Looks good to me!

[wolfv]

Awesome job @borchero - there is actually one other really nice use case for this. And that's "caching" the activation. We have a way to cache a task based on the environment + inputs + outputs.

The idea would be that we hash the current environment, and store the activation variables that we generated as a JSON file. When the environment stays the same, the activation should then subsequently be much faster.

Are you perhaps interested in implementing that? :) #973

I am not sure if we should special-case the $PATH variable in that activation-cache scenario, though. And wether the current set of environment variables should be part of the hash or not.

[borchero]

Sorry for the late reply @wolfv, I missed this comment before!

Sounds like a very reasonable feature, I'll have a look soonish 😁