Investigate feasibility of parser backtracking for ambiguous options

[remkop]

Follow-up on #979, #981, #370, #659

Any time an option has a range of parameters, like @Option(arity = "0..1") or @Option(arity = "1..*"), and the command also defines one or more positional parameters, there is ambiguity whether a command line parameter should be considered an option parameter or a positional parameter.

Sometimes it is possible to resolve the ambiguity locally, and picocli currently applies heuristics to determine if the value is a subcommand or (likely) another option.

In other cases, it may possible to resolve the ambiguity globally: if a command has an option with an optional parameter, and a required positional parameter, then interpreting the parameter following the option (--option value) as a positional parameter is the only interpretation that does not result in an error.

This requires the parser "marking" points of ambiguity and processing the input multiple times, using different interpretations of the "grammar" defined by the command options and parameters.

[remkop]

I don't see myself implementing this.

Picocli's parser plugins offer an easier alternative.