Missing tests for floating point with integral > 2^31 -1

[jrahlf]

While working on #84, I noticed that the test coverage for large floating point values seems insufficient.
When changing the integral field type to int32_t, the test suite still passes. This makes me believe that there is no test case for floating point values with the integral part > 2^31 -1.

printf/src/printf/printf.c

Line 423 in 36ad016

int_fast64_t integral;

[eyalroz]

I'll have a look.

[eyalroz]

@jrahlf : How's this?

[jrahlf]

Now fails for 32-bit integral type as it should.

[eyalroz]

@jrahlf : Hmm... should it, though? Or - do you mean, on the version you're working on, which uses 32-bit-integers only, internally?

[jrahlf]

What I meant: if you change the type from uint_fast64_t to uint32_t, the test suite fails.

[eyalroz]

I think we're ok now.