Optimize bound checks and limit goroutines to GOMAXPROCS(0).

[charlievieth]

Optimizing the conversion function bound checks increased overall performance by 14.8%. The benchmarks were also updated to better reflect conversion performance.

Also, runtime.GOMAXPROCS(0) is used instead of runtime.NumCPU() to determine the number of spawned goroutines. This allows clients to limit the number of spawned goroutines, and slightly increases performance.

Test Results

go test -cpu 1 -bench .

Test	Previous	Updated	Change
Benchmark_Nearest_RGBA	3588122	3079371	16.5%
Benchmark_Bilinear_RGBA	5719682	4835984	18.3%
Benchmark_Bicubic_RGBA	7914568	6702097	18.1%
Benchmark_Mitchell_RGBA	8866317	8063793	10.0%
Benchmark_Lanczos2_RGBA	8049316	6853025	17.5%
Benchmark_Lanczos3_RGBA	10525981	8845505	19.0%
Benchmark_Nearest_YCC	4120125	3766400	9.4%
Benchmark_Bilinear_YCC	5673952	5048623	12.4%
Benchmark_Bicubic_YCC	7646177	6699744	14.1%
Benchmark_Mitchell_YCC	8558137	7531485	13.6%
Benchmark_Lanczos2_YCC	7738619	6814587	13.6%
Benchmark_Lanczos3_YCC	9711631	8500676	14.2%
Total			14.8%

[nfnt]

Thank you!