ASoC: msm: Add MultiKeyword support for Listen

Listen is a Voice activation and recognition feature.
Currently Listen feature supports recognition of single
keyword for a given session. This change adds support
for MultiKeyword in Listen interface header. Userspace
hal will use this interface header to determine Sound
Model version based on App ID.

Change-Id: I5fb77297121ecc52eeeabbd522f536b97c73dc8f
Signed-off-by: Venkat Sudhir <vsudhir@codeaurora.org>