enhancement(adr-dir): make finding the adr-dir more intuitive

[joshrotenberg]

adr-tools creates the .adr-dir file on init and then uses that for subsequent runs to decide where to put/get ADRs. This works fine, and I've reproduced this behavior in adrs, but I think it would be cleaner to use a global flag tied to an environment variable, as well as just use the docs/adr as a default everywhere.

The flow would be something like this:

• If no .adr-dir exists, and no docs/adr exists, default to docs/adr without writing a .adr-dir file
• If the global --directory flag is set for any command, use that value for all operations, even if a .adr-dir exists.
• If the ADR_DIRECTORY environment variable is set, use that value for all operations even if a .adr-dir exists.
• If no .adr-dir exists but the docs/adr directory does exist, use the docs/adr directory.

I think this covers most potential situations while maintaining backwards compatibility and keeping sane defaults.

Steps off the top of my head:

1. Add the directory flag in main.rs as a global flag, with the env marker to also pull it from the environment.
2. Process the flag before calling the subcommand, and pass the directory value to any subcommand as a Path (or maybe PathBuf?).
3. Change anything that uses .adr-dir now to always use that value.

[joshrotenberg]

Also, it probably makes sense to actually write ADRs for things that diverge from adr-tools functionality since this is, like, a tool for ADRs.