dsp: lsm: remove the size checking for sound model

Currently q6lsm driver only allows buffer to be allocated for
SVA sound model with size larger than 512 byte in function
q6lsm_snd_model_buf_alloc.This causes issue when a valid SVA
sound model is smaller than 512 byte. Since there is no
restriction in ADSP for sound model size, remove this size
checking to allow sound models with smaller size.

Change-Id: I716dc8c0e1a4e3189cd334ef692ff07f3e4635db
Signed-off-by: Xiaoyu Ye <benyxy@codeaurora.org>