Merge series from Bard Liao <yung-chuan.liao@linux.intel.com>:
Rather than keeping a single array of CPU dai link components allocate a
smaller one for each DAI link, this reduces the amount of state that
needs to be passed back and forth in the driver.