Standardise logic for failing/warning when an error occurs during terraform version detection

[MatthewJohn]

Is your feature request related to a problem? Please describe.

There are two use-cases when running tfswitch:

• As a local user, I wish to run tfswitch. If tfswitch runs into a problem (i.e. it cannot parse my Terraform or it thinks there's an issue with a terragrunt config file), I would like it to show a warning and continue as normal (perhaps I specified the version as an arg, or have the version defined elsewhere).
• As a CI/CD execution server, if I am expected to parse Terraform and obtain a version, if the terraform cannot be parsed, I do not wish to begin an endless loop of waiting for user input (because there won't ever be any)

Describe the solution you'd like

As suggested in the thread on/after https://github.com/warrensbox/terraform-switcher/pull/429#discussion_r1624655080, we could/should detect for a valid TTY and act accordingly. If there is no TTY, then we should fail when these errors occur, otherwise, we should continue to ask the user for input.

[yermulnik]

Out of curiosity: what that screencast (git switch branch + open main.go in vi) is meant for? 🤔

[MatthewJohn]

Out of curiosity: what that screencast (git switch branch + open main.go in vi) is meant for? 🤔

Ah, I saw it in the new issue template and wondered what it was - and then forgot to remove it!

[warrensbox]

Can this wait after release 1.2.0?

[yermulnik]

Can this wait after release 1.2.0?

It probably can. Though this "bug" may impact someone's CI/CD pipeline by halting it to wait for an input w/o a meaningful reason. I'd personally target this to be release as soon as possible. Though! Though we may (and need) to establish release cadence if we may (and can) provide a more-or-less stable release cadence I guess.