Wrap math.MaxInt64 in int64() for 32-bit cross-compilation

[jacohend]

We must wrap math.MaxInt64 calls in int64() for rpi/arm/32-bit support, otherwise cross-compilation yields this error:

cmd/tarocli/addrs.go:144:9: cannot use math.MaxInt64 (untyped int constant 9223372036854775807) as int value in assignment (overflows)

[guggero]

Thanks for the fix. Though I'm not sure we should explicitly support a 32-bit architecture. With lnd this lead to problems before with some limitations around max DB size. What architecture/device are you targeting?

[jacohend]

I see, the smaller 32 bit max make sense as a problem for db limits.

What architecture/device are you targeting?

Arm 6 and 7, raspbian.