I've been on a reprocessing spree and it's finally arrived at the Tulip. Here's what I did:

H, O and S masters:
dynamic crop
deconvolution
starXterminator (extracting stars)
histogram transformation
tgv denoise
clone the ha as lum (the S and O were too noisy to use in the luminance)

HOS
channel combination (HOS)
histogram transformation (align peaks & touch up black point)
channel combination in the lum
pixel math in the stars

lum:
LHE (2 scales)
MLT sharpening

stars
histogram transformation

Below is the original text and processing for revision A.
-------------------------------------------------

Among the bountiful jewels of Cygnus is Sh2-101, the Tulip Nebula.  This is an HOS palette version that turned out to be a lot tougher to process than anticipated, especially when you factor in the amount of data we collected. Unfortunately, the smoke from the fires out west has begun to impact the imaging at SRO. The O and S were the most impacted, especially the O. Realistically we could have thrown away at least ¾ of the data we collected and not hurt our ultimate image quality.

Because the S and particularly the O was so thin, it was tough to stretch without introducing too much noise. It took a few tries to find the right balance and I had to err on the side of controlling the noise. I was really looking forward to this target as I've seen some spectacular images of it but I think we are going to need to revisit it next year with hopefully no smoke in the sky. Even so, I was happy to get this result out of it even it made me feel like I was processing my early attempts from a couple of years ago. It was a humbling experience.

I decided on HOS because it seemed to fit the nebula's namesake, a tulip. I was tempted to go with a straight SHO...the green worked pretty well here but ultimately decided I liked the HOS better.

Here's how I ultimately tackled it:
Preprocessing with WBPP and the Normalize Scale Gradient scripts. NSG recommended we throw away nearly all the data which we tried and also keeping everything. It didn't look like keeping everything was hurting so I decided to process it further.

H-a, O and S:
dynamic crop
deconvolution
MLT noise reduction
histogram transformation
TGV denoise (O and S only)

HOS:
dhannel dcombination (HOS)
background neutralization
histogram transformation
star de-emphasis script
dark structure enhance script
local histogram equalization (two scales)
scnr (using star mask to get just the largest stars that had green fringes)
unsharp mask (very mild)
dynamic crop to final dimensions

By splitting up the histogram transformation into two steps, I was able to better control the noise because the channels were stretched equally in the second stretch where the were stretched individually in the first stretch. If I tackled it in one stretch the O noise became objectionable though the resulting color contrast was nice. This felt like the best compromise even though it was not quite what I set out to achieve. Or to put is a friend is fond of saying, "if you can't get what you like you are forced to like what you can get. "