Add precision on tests using toBeCloseTo.

[winksaville]

According to this the default precision is 2 that seems to few, so using 10.

Also, added the precision test to validate the test.

[MaxGraey]

Better using 5. Because machine epsilon for float is 1.19e-07, but last 2 digits contain rounded error

[winksaville]

I'll change it but I'm curious, Is 1.19e-07 for float32 I'd have to believe float64 is larger. Do you have a link for the wasm machine spec?

[winksaville]

So on wikipedia I see float32 has 7.22 decimal digits of and float64 has 15.95 should we assume float32 or taylor the test depending on float32 or float64?

[MaxGraey]

In current test we just validate import / export routines, so even 2 digits enough. For simplicity and perfection we can use 5 digits. This will cover as float and double. Do not complicate such a simple test.

[winksaville]

If two is enough, then no need to merge, simple is better.

[nidin]

closing, we can add tight precision check when testing stdlib implementation.