fix multiple zeros on split_thousand

[Laifsyn]

Related issue: #11

Because the decimals weren't being truncated, while !num.is_zero() { never evaluates to 0 until the BigFloat can no longer display the super small number.

For the curious ones, here's a benchmark that could display the small performance improvement (Using Criterion)

Before fix

Num2Words to_cardinal()

	EN	FR	FR_BE	FR_CH	UK
hundreds	24.46 us (✅ 1.00x)	24.99 us (✅ 1.02x slower)	24.57 us (✅ 1.00x slower)	24.46 us (✅ 1.00x slower)	25.06 us (✅ 1.02x slower)
thousands	25.90 us (✅ 1.00x)	25.84 us (✅ 1.00x faster)	26.01 us (✅ 1.00x slower)	25.95 us (✅ 1.00x slower)	26.64 us (✅ 1.03x slower)
millions	26.96 us (✅ 1.00x)	26.99 us (✅ 1.00x slower)	27.39 us (✅ 1.02x slower)	27.31 us (✅ 1.01x slower)	27.88 us (✅ 1.03x slower)
billions	28.18 us (✅ 1.00x)	28.28 us (✅ 1.00x slower)	28.38 us (✅ 1.01x slower)	28.38 us (✅ 1.01x slower)	29.05 us (✅ 1.03x slower)
trillions	29.08 us (✅ 1.00x)	29.42 us (✅ 1.01x slower)	29.37 us (✅ 1.01x slower)	29.31 us (✅ 1.01x slower)	30.27 us (✅ 1.04x slower)
quadrillions	30.67 us (✅ 1.00x)	30.89 us (✅ 1.01x slower)	31.16 us (✅ 1.02x slower)	31.88 us (✅ 1.04x slower)	31.21 us (✅ 1.02x slower)
nonillion	37.69 us (✅ 1.00x)	37.98 us (✅ 1.01x slower)	37.16 us (✅ 1.01x faster)	37.15 us (✅ 1.01x faster)	38.37 us (✅ 1.02x slower)

After the fix

Num2Words to_cardinal()

	EN	FR	FR_BE	FR_CH	UK
hundreds	613.99 ns (✅ 1.00x)	574.91 ns (✅ 1.07x faster)	578.37 ns (✅ 1.06x faster)	592.72 ns (✅ 1.04x faster)	535.31 ns (✅ 1.15x faster)
thousands	1.79 us (✅ 1.00x)	1.70 us (✅ 1.05x faster)	1.70 us (✅ 1.05x faster)	1.68 us (✅ 1.07x faster)	1.68 us (✅ 1.07x faster)
millions	2.99 us (✅ 1.00x)	2.81 us (✅ 1.06x faster)	2.82 us (✅ 1.06x faster)	2.82 us (✅ 1.06x faster)	2.93 us (✅ 1.02x faster)
billions	4.12 us (✅ 1.00x)	4.02 us (✅ 1.02x faster)	4.05 us (✅ 1.02x faster)	4.00 us (✅ 1.03x faster)	3.90 us (✅ 1.06x faster)
trillions	5.12 us (✅ 1.00x)	5.23 us (✅ 1.02x slower)	5.00 us (✅ 1.02x faster)	5.21 us (✅ 1.02x slower)	5.15 us (✅ 1.01x slower)
quadrillions	6.80 us (✅ 1.00x)	6.66 us (✅ 1.02x faster)	6.36 us (✅ 1.07x faster)	6.61 us (✅ 1.03x faster)	6.19 us (✅ 1.10x faster)
nonillion	13.71 us (✅ 1.00x)	13.81 us (✅ 1.01x slower)	13.80 us (✅ 1.01x slower)	13.78 us (✅ 1.01x slower)	14.36 us (✅ 1.05x slower)

Made with criterion-table