oh oh , i cannot agree with your first comment. even power planes do generate emmisions. how can i explain this ?
the field that goes from vcc to gnd needs the gnd plane, but at the edges the field does not find enough gnd plane so it starts to emit at the edges of the power plane corners.
again here is where you can save up to 3dB without any extra cost if you do a proper power plane layout.
yes, a 4layer is about 1.3 times the cost of an 2layer board. but the benifits you get for power stability and emc is enormously.
yes this would be a replacement for a lot of parts you would need on a 2layer board and comparing prices (including manufatoring ...) the 4Layer will win in most cases.

in your design (no external bus) a lot of things are very easy.
i would place an L-C lowpassfilter on all signals leaving / entering the boards. the cap goes to gnd. both placed as close to the connector as possible or if you place an metal box on the pcb, place them inside the box.
make shure that clock signals (all high speed signals) have as little as possible layer changes. in best case are only on one layer.
another easy winner is, that you place the supply pins traces with vias on the inner side of the device to reduce the loop between them and so minimize the effective area of this single loop antenna. what is the only way to reduce the emission in this case.
yes the dc filter from murata can help. but some of them can be made much better if you place another cap at both sides of this filter as close as possible.

chassis grounded ? why does everybody thing that if you "ground" something then there is no emission ?
gnd is just a signal and gnd can emit noise as well. i prefer the other way, i reduce emision where it is produced so the less i have the less can make noise somewhere else.

michael