Remove shell

[petersutter]

What this PR does / why we need it:
With gardener/gardenctl-v2#86 it is now easier to target a (shoot) cluster. Running eval $(gardenctl kubectl-env bash) after running the target command is not required any more to set the KUBECONFIG env variable. Instead, it only has to be run once (e.g. by including it into the shell profile). See more details here https://github.com/gardener/gardenctl-v2/blob/master/docs/concepts/cli_interaction.md#behavior-with-symlink

Which issue(s) this PR fixes:
Fixes #

Special notes for your reviewer:

Release note:

The `gardenctl-v2` commands that can be copied from the cluster details page to target the "Cluster" or the "Control Plane" will not include the command to set the `KUBECONFIG` environment variable anymore (e.g. `eval $(gardenctl kubectl-env bash)` in case of `bash`)
- A warning message is displayed when running the `target` command in case the `KUBECONFIG` environment variable does not point to the current target
- The [`rc` command](https://github.com/gardener/gardenctl-v2/blob/master/docs/help/gardenctl_rc.md) startup script sets the `KUBECONFIG` variable when evaluating the command output, usually as part of the shell profile, e.g. `source <(gardenctl rc bash)` in case of `bash`. Requires `gardenctl` `v2.1.0`
- If you do not load the `gardenctl` `rc` command in your shell profile or do not have the `gk` or, with `bash` as an example,  `eval "$(gardenctl kubectl-env bash)"` in your `bash` profile, you will have to run it manually in order to have the `KUBECONFIG` point to the kubeconfig of the current target

[holgerkoser]

/lgtm

[grolu]

/lgtm

[petersutter]

/hold until we have adapted the rc command in gardenctl to also eval the kubectl-env command

[gardener-robot]

@petersutter You need rebase this pull request with latest master branch. Please check.

[petersutter]

/unhold