No, that is a good and valid design approach. It also allows to test your device with 50 Ohm test equipment, independent of the load + matching network.
Choose the Filter's Input Output Real Impedances carefully regarding to those impedances to be matched. Choose appropriate values so that matching circuit won't consist of weird elements..
Here is a fictitious example for illustration: Load impedance: 1 Ω + 2.55 nH
Matching frequency: 500 MHz
Source impedance: 2 Ω / 6 Ω / 20 Ω (selectable within certain limits)
Matching made with a low "pass structure" (n=4). As you can see, there is no well defined passband.
frdm90 didn't yet release a matching bandwidth specification. I presume, it's not possible to achieve a higher bandwidth with LC network for the example matching problem.
You cannot match over this bandwidth with simple few matching circuit.Load Impedance is not constant and varies by frequency..
Is your intention to match these termination impedances ( both source and load ) over the bandwidth..?
Simple matching circuit wouldn't work.. Impedance Matching Transformer ( not exact but approximate) may be a solution but matching is not guaranteed at lower frequencies.