Implement download subcommand, optional positional model name argument #348

Job	Run time
gather-models-gpu	2s
compile-gguf (macos-14)	4m 29s
runner-et (macos-14-xlarge)	10m 26s
gather-models-cpu	6s
test-gguf-util (macos-14)	7m 12s
test-tinystories-eager (macos-12)	9m 43s
test-tinystories-executorch (32-core-ubuntu)	15m 20s
torchchat-command-load-test (macos-14)	1m 1s
test-mps / macos-job	2m 7s
test-mps-dtype / macos-job	3m 31s
test-gpu-compile (cuda, stories15M) / linux-job	17m 52s
test-gpu-aoti (cuda, stories15M) / linux-job	22m 35s
test-cpu-aoti (x86_64, stories15M)	14m 16s
test-cpu-compile (x86_64, stories15M)	11m 6s
test-runner-aoti (x86_64, stories15M)	2m 6s
test-cpu-compile (aarch64, stories15M)	6m 33s
test-runner-aoti (aarch64, stories15M)	1m 18s
test-cpu-aoti (aarch64, stories15M)	7m 25s
	2h 17m 8s

Provide feedback