Extras handling refactor

[henryiii]

In the interests of solving #36, this PR refactors the way extras are handled. The implementation:

• parse no longer returns a vector (not well supported anyway, for example with CLI11_PARSE
• Added remaining() and remaining_size(), which provide access into the unparsed options. They support a "recurse" bool flag that causes them to collect all subcommand arguments, too. They are not "reversed".
• allow_extras and prefix_command are now valid on subcommands, and inherit the setting from the parent upon creation (like the other options)

Oddities that remain:

• The vector form of parse still expects a "reversed" vector; it updates itself with the remaining options, however. Reordering might be unintentionally possible with multiple subcommands.

Will need to maybe add more tests, and needs docs and changelog.

[codecov]

Codecov Report

Merging #37 into master will not change coverage.
The diff coverage is 100%.

@@          Coverage Diff          @@
##           master    #37   +/-   ##
=====================================
  Coverage     100%   100%           
=====================================
  Files           9      9           
  Lines        1141   1149    +8     
  Branches      233    237    +4     
=====================================
+ Hits         1141   1149    +8

Impacted Files	Coverage Δ
include/CLI/Option.hpp	100% <ø> (ø)	⬆️
include/CLI/App.hpp	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3de7832...e94d465. Read the comment docs.

[henryiii]

@lambdafu , let me know how this works (and maybe the other two PRs, too) :)

[lambdafu]

Looks good! I'll implement this in a larger program and see if there are any issues remaining. Thanks!