I question the bit about insertion loss 7MHz away being a non-issue
with only a single open stub across the line. If you put an open stub
of lossless 50 ohm line a quarter-wave long at 153MHz across the feed
line, it puts a short across the line at 153, but it puts 0.26-j3.58
ohms across it at 146MHz. That's a 99% reflection, or SWR close to
200:1 at 146. It's only about a 17dB loss, but that's more loss than
I'd want.

But if you also put at the same point across the line either a shorted
stub or an inductor to resonate that, it will let 146 pass much more
easily...the inductance required is only about 4nH, though, so it ends
up being a pretty short piece of wire, and a Q of 20 results in almost
3dB loss. Still, that's a lot better than 17dB loss.

Also, you can tune a stub by making it very slightly shorter than
calculated and adding capacitance to the open end--an air-dielectric 50
ohm stub cut 1cm short, with a 1pF capacitance across the open end,
tunes back to 153MHz. A little piston cap lets you tune it right on
the offending signal. (You can also make the 146MHz peak tuneable by
using a stub a bit _longer_ than 1/4 wave at 146 in place of the
inductor, and putting a tuning cap on it too...then it's easy to tweak
right in for optimum performance.)

These what-if scenarios are easy to simulate in RFSim99, which is
freeware. There are lots of variations on the theme, but at very
least, it's a quick check of a proposed solution to see if it's really
all that good at both getting rid of the offender and passing the
desired signal. Be sure to include expected loss resistance to
simulate the finite Q of real components.

Cheers,
Tom