Generic topology to support different sample rate/format input and output.

[keyonjie]

We need to switch topologies to support different sample rate/format input and output, but this is unacceptable for the end-user.

We need to provide a generic topology file that it does support all sample rate/format(be inconsistent with the capability we are claiming).

This may need architecture enhancement in both SOFT and SOF.

[lgirdwood]

@keyonjie care to give example of what you mean.

[keyonjie]

@lgirdwood the most significant points I can think of is:

1. support to allocate/reallocate FW buffers at hw_params() time, now it is done at pipeline creation stage;
2. don't need to specify the usage of volume and src components in pipeline definition. The easier solution ATM is that these two kinds of components are added by default, and do only exactly copying when no sample format/rate change needed.

[lgirdwood]

@keyonjie 1. we pre-allocate buffers like ALSA to keep things simple. If the user needs a bigger buffer then they ask for it in topology.
2. Nak, pipeline definitions should be exactly what the user needs and not contain extra processing that is not needed. If user want SRC/format conversion then this must be defined in thier pipeline. There is nothing stopping us defining a template for this pipeline (like pasthrough) that can be included by users defining their own pipelines.