shebang sub-command naming

[Gedochao]

Can we do better than shebang on the name? It seems like obscure jargon to me

(answering #1740 (comment))
@SethTisue any suggestions for an alias?

I actually don't think it's that obscure of a jargon...
The current shebang sub-command has 2 main use cases:

• to be used in the shebang header, as per the Unix standard, for which the name seems appropriate;
• to be used from the command line with legacy syntax backwards compatible with the old scala runner.

Although I believe we'll be keeping the shebang name, as the former usecase is the main one, I think we could potentially have an alias to make it easier to guess the second one... provided that we can think of a good name for it, of course.
Would something like run-legacy do the trick? feels verbose to me...

Also, I wonder if an alias is even necessary... if a user cares about backwards compatibility in this context when migrating from the old scala runner to Scala CLI, then it's probably fair to assume he's a rare specimen of a power user. Which means he should probably be okay with obscurely named sub-commands...

We don't want to promote using shebang from the command line for regular users, they should just learn the run syntax.

[Flowdalic]

Does it have to be a subcommand? Can't scala-cli not inspect the name under it was called and act accordingly? Then you could have e.g.,

#!/usr/bin/env scala-script

which looks nice, clean and idiomatic. And /usr/bin/scala-script is a symlink to /usr/bin/scala-cli.

[Gedochao]

That is actually how we've been differentiating between SIP-mode and the standard scala-cli... well, we still do, until #1824 is resolved (a solution is already in-progress in #1849)

Regardless, I believe we do still want shebang available as a sub-command, as some people prefer its command line behaviour over run. Not to mention it's easier to document it as a sub-command.
shebang sort of mimics how the old scala runner treated program arguments (all args started with the second are treated as program args rather than inputs, no need for --) and apparently some people are used to handling it this way.