Add --remote option to endpoint delete

[LeiGlobus]

This adds functionality to delete endpoints when a local config is not present.

It differs from the original story in that this no longer uses t he --remote option, but just as part of normal delete, with the usual --yes and --force options behaving as they should intuitively.

A few notes:

• Modified get_ep_dir_by_name_or_uuid() to take an extra parameter with default so it doesn't raise if an UUID is specified that does not exist on the system, and also to pass the uuid on as an argument
• Added a new wrapper name_or_uuid_arg_non_local that takes advantage of the above without affecting the existing name_or_uuid_arg
• Note that the happy deletion path calls the Globus Compute Client's delete_endpoint() which has the same name but is a direct API call via web_client.delete_endpoint()

[joshbryan-globus]

[sc-31206]

[joshbryan-globus]

[sc-32452]

[rjmello]

This looks good, but I think a simpler, more intuitive path forward is to refactor the --force flag to attempt to delete the endpoint locally and remotely, pushing through exceptions for each operation. If the endpoint cannot be found locally, then we would try to delete it remotely with a provided UUID. If the user does not provide a UUID, we can raise an informative exception that asks them to do so.

Another path might be to drop the --force flag in favor of --local and --remote flags. By default, we would attempt to perform both operations, but if the user specifies one of these flags, then we would only attempt the corresponding operation.

[rjmello]

We should probably move this to the top of the function and out of the if remote: block so that we always give a warning.

[LeiGlobus]

I agree that it's worthwhile moving the confirmation to the top and changing existing logic a little.

[LeiGlobus]

This looks good, but I think a simpler, more intuitive path forward is to refactor the --force flag to attempt to delete the endpoint locally and remotely, pushing through exceptions for each operation. If the endpoint cannot be found locally, then we would try to delete it remotely with a provided UUID. If the user does not provide a UUID, we can raise an informative exception that asks them to do so.

Another path might be to drop the --force flag in favor of --local and --remote flags. By default, we would attempt to perform both operations, but if the user specifies one of these flags, then we would only attempt the corresponding operation.

I did think about doing something similar but went with coding exactly as the story dictated, figuring that to deviate would warrant a scrum/dev thread instead of just changing the behavior between me and the reviewer.

I will summarize the current and other (your) path in a dev thread to gather opinions, potentially to be discussed as a ramp item tomorrow afternoon if it isn't clearly resovled before.