Prevent fork nightmare with PAGER=batcat

[johnmatthiggins]

Used rsamuelklatchko's suggested changes to fix recursive fork issue. Fixes issue #2023.

[Enselic]

Thanks for working on this! I added some comments.

[johnmatthiggins]

Made commits with suggested changes, ready for further review.

[Enselic]

Thanks! Much clearer code. I proposed one possible further simplification. Can you also please add an entry to CHANGELOG.md? Like this:

## Bugfixes

- Prevent fork nightmare with `PAGER=batcat`. See #2235 (@johnmatthiggins)

When that is done I think I can set this PR as Approved!

[johnmatthiggins]

Added to change log and altered match expression.

[Enselic]

I forgot to comment on the arg0 name, so I pushed a commit for that. I hope you didn't mind.

This looks good to me now! Thanks a lot.

As far as I can tell our current regression tests cover this PR. For example, if we hardcode is_current_bin_pager to either true or false, tests start failing.

I didn't expect it to be, but just to make sure I ensured that startup time is not in practice affected, and it wasn't:

% hyperfine --export-markdown /dev/tty -L bat ./target/release/bat,bat-master-c8b1187 '{bat} -f --paging=always --theme="Monokai Extended" examples/simple.rs'

Command	Mean [ms]	Min [ms]	Max [ms]	Relative
./target/release/bat -f --paging=always --theme="Monokai Extended" examples/simple.rs	11.0 ± 1.1	10.1	21.2	1.00
bat-master-c8b1187 -f --paging=always --theme="Monokai Extended" examples/simple.rs	11.1 ± 1.1	10.1	21.9	1.01 ± 0.14

[Enselic]

About time to merge this now I think.