Gordon, Howard,

With all the parallel thread it was somehow missed.

FYI memory says both capacitance and inductance change.

Looking at the idea.  I think the basis is sound.  It also may improve the trace
to trace coupling (less coupling).  I worry that the increased currents may
highlight other board level RF current loops.  The output transformer has
to be rewound and that in itself could be an improvement. The ouch is that
converting back to 50 ohms and in a spot before K3 (TR) to not impact RX.
Also transformers have loss and bulk.

The current are likely still low enough and the traces may be light but I think 
the worst that can come of it is some unwanted likely small loss.  They will
not burn up but it is always a risk running lower impedance as the I^2R
losses go up for conductors.  I can't easily evaluate as its RF and skin
effects and wire diameter are are higher and messy to calculate plus it
would have a lot of not validated assumptions.  Its faster to test than guess.

If someone can try it,  they only need to do the 80m filter, wind two transformers
to see and test...

FYI ALL:
I added a image NF1.jpg its a capture from the Rigol on the 6th showing the
blow by of the relays from where the power amp input to the filter selection
relays to the antenna terminal in receive mode. no relays active.  So what
your seeing going past the 10m filter to the antenna jack via layout and
open TR relay (K3)  contacts.   The markers in the table are set to the
3db cutoff or close for a reference point.  It translates to if the filters
are perfect the is as good as it gets due to wires and relays.  What
would I like to see for that case? better than -60db.  The only place
it gets there is 1mhz (extreme left edge).  The source signal is 0dbm
or 1mW.