New tests feature + benchmarks

[tdehoff]

1. Selected tests to run can be specified with argv. Example:

(MojoBLAS) tdehoff@hudson:~/MojoBLAS$ mojo test-level1.mojo dot asum iamax
--- MojoBLAS Level 1 routines testing ---

Running 3 tests for /home/tdehoff/MojoBLAS/test-level1.mojo 
    PASS [ 781.155 ] test_dot
    PASS [ 3.285 ] test_asum
    PASS [ 3.285 ] test_iamax
--------
Summary [ 787.726 ] 3 tests run: 3 passed , 0 failed , 0 skipped 

2. Benchmarks take argvs too: can specify DType, problem size, num iterations. I used BLAS++ tests as a reference. Example:

(MojoBLAS) tdehoff@hudson:~/MojoBLAS$ mojo bench-level1.mojo --type float32 --n 10000
op, device, dtype, n, iters, avg time, bandwidth
copy, NVIDIA H100 NVL , float32 , 10000 , 100 , 7375 ns, 10.846486958455243 GB/s

To accommodate the updates, I had to change the generate_random_arr signature.
Feel free to suggest any improvements in what and how we print, as well as more argv-controlled parameters. I'll add benchmarks for the remaining routines once I receive some feedback.

[jacksonmowry]

Sweet! Thank you so much

[acandershock]

looks good! the huge if else statement in test-level1.mojo reminded me of gregor's class, lol