Allow specifying the default domain for unbounded integer variables in flatzinc interface

[CervEdin]

If you write a minizinc model with 4095+ var int the model will be declared invalid

array[1..4096] of var int: a;

because the domain defaults to -(2^50)..(2^50)
https://github.com/google/or-tools/blob/ed94162b910fa58896db99191378d3b71a5313af/ortools/flatzinc/cp_model_fz_solver.cc#L45C1-L47C1
and this will fail the model overflow check.

It would be useful to allow users to adjust this default, say by passing a CLI parameter to override this default value.
I guess the canonical way of doing this is to add another extraFlags

or-tools/ortools/flatzinc/cp-sat.msc.in

Line 11 in ed94162

["--params", "Provide parameters interpreted as a text SatParameters proto", "string", ""]

so something like

"extraFlags": [
    ["--params", "Provide parameters interpreted as a text SatParameters proto", "string", ""],
    ["--unbounded-int-var-domain", "Domain of unbounded int variables", "int", ""]
  ],

[lperron]

already there:
ABSL_FLAG(int64_t, fz_int_max, int64_t{1} << 50,
"Default max value for unbounded integer variables.");

[lperron]

Implemented