I have two claims regarding input matching in PA.

1) The PAE does not  depend on the input matching network , its only the power gain that depends on it. The PAE of a input mismatched PA will be same as the matched one..

2)The input matching s parameters depends on the Ropt --> 50 ohms transformation network and is independent of the harmonic supression network on the output of the device..Since Ropt -->50ohms network is tuned at fo ..

Am i thinking in the right direction?

PAE is the difference between the output and input power levels devided by the dc power..the main role of input network is to provide maximum power from soruce to the gate of the device ...The differnce  remains the same even for mismatch at the input and hecne the PAE must remain same..

The output  network of linear PA consists for harmonic filter and load transformation network..At  fo  the parallel harmonic filter is open (or it disappears) and the network   "consists" of device and load  transformation network..So , the s parameters depends on the load transformation network provided the device is not altered..

I think the bigger issue regarding input matching is that PAE is defined as the power delivered to the input and not the available power of the input source.

Under this assumption, you must match the input and attempt to deliver as much output power as possible to the load at the fundamental. Ropt in this case is matching the output to deliver as much output power as possible, but there is still the assumption that the input is power matched for the PAE calculation/measurement. If you mismatch the input then Pout-Pin will suffer.

Steve Cripps gives a fairly clear account of this in his book entitled RF Power Amplifiers for Wireless Communications, 2nd Edition published by Artech House. I recommend it as a good reference on RF power amplifier fundamentals.