You have mitigated a contributing problem. It may have been the more important one. Actually probably was.

You still have the other problem. They were acting in unison. The changes with ambient temperatures you reported. Was an astute observation. I would tend to label it as an effect of the fuel temperature and ignore it. Rather than an expansion or contraction of a hose or sealing washer. In the past people possibly never considered this if the air was being pulled in.

The vacuum pulled by the negative portions of the spike pulses just might vary. With the fuels changing viscosity with temperature to some extent. By elevating the fuel pressure you have also established a higher baseline for those pulses. You may not be seeing as much air as the pulses may be producing a lesser vacuum component.

There are many ways to locate the still existing air leak.. First I would concentrate on the stem area seals of the secondary fuel filter. Replace them if cheap. Plus check for cracks in the housing. The cracks are not all that common but not unknown as well. Then you can go higher tech or lower.

Lower is basically remove the banjo bolts and replace the sealing washers. Or heat them red hot if they still appear in decent condition and let them air cool. This annealing removes any work hardening by softening them if they are copper. Do not over torque the hollow banjo bolts.

Higher tech is using a can of cold spray. Spraying a banjo bolt will shrink the fitting a little. Compressing the sealing washers a little more. Frost will build up as well. This frost will also act to reduce the amount of air the fitting can take in.

You just watch for air reduction or signifigant change in the relief valves output. It may take a little time and keeping that fitting really cold to see a difference in air output.

You can easily verify the test if you find something. The relief valve banjo bolt can also be a point of air ingress so test it as well. Remember as you freeze the fitting the attached hose may be the culprit. except for the relief valves hose end. You will be shrinking the hose end as well with the cold spray. When addressing banjo bolts this way.

If the leak is more obscure and it probably is not. There are other simple tests to locate it. Or to determine you have not missed it in the previously tested area. For example running a hose around the whole secondary fuel fixture and filter. Obviously if the problem is gone it is there somewhere. Problem remains it is not of course. A short duration run without the secondary filter system in the loop is not a big issue.

If you see air you do not want to. If some assumptions that your ingress of air is temperature sensitive. Perhaps a lot of the others over the years where as well.

This was just not noticed perhaps. Careful use of say a heat gun may locate them fast as well if that is the case. Air leakages into the fuel system are very common failures. The majority get resolved with no great production. .