In the trigger trees, L2sum is a 64 bit stream of trigger level 2 outputs, taking into account prescaling; it is stored as an array of two C++ ints. The upper int, L2sum[1] appears to be a copy of the lower int, L2sum[0] and thus only the L2sum[0] half is valid for selecting triggers. Fortunately, all the triggers which are supposed to be in L2sum[1] are takeall and are available in the FP201-->TCU inputs bitstream, stored in trigger trees as lastdsm[5]. Note that lastdsm[5] does not account for prescaling and is not valid for selecting prescaled triggers on events.

The following table summarises all of the triggers used. The non-obvious columns are:
idx -- local trigger index for my code
DSM -- dsm board name (all FP201 here)
chan -- TCU input channel (all 5 here)
bit -- TCU input bit for this trigger
which_branch -- whether to use L2sum[0] or lastdsm when selecting this trigger (special case is for "All", see below)

The All trigger, although a bit of a misnomer, is the OR of all of the above triggers; it uses both L2sum[0] and lastdsm[5] via the "allmasks" which are listed at the bottom of the table; basically, All trigger is marked as "fired" if any of the other triggers fired.

Below the table, the region with the black background is information for a single event, with it's L2sum, lastdsm[5], and which triggers fired.

.
.
.
.

This is a matrix of the overlap between these triggers; the main diagonal represents the total number a particular trigger fired; all other numbers indicate the number of times the trigger named in its row and column fired in the same event.

.
.
.
.

Here is a table of images for plots of kinematic variables against one another.
-- Three event classes are shown, occupying 12 rows each: single photon clusters, pions, and >2 photon clusters
-- each row is for a different pair (or single distribution) of variables
-- each column is for a different trigger
-- I have these plots available for each run, someone may make gifs or videos out of them to show how they evolve throughout the run
.
.
.
.

I've looked at pT, E, Eta, and Phi distributions as a function of run number for each of these triggers. The following PDFs show these plots:
-- one trigger per page
-- top plot is kinematic variable vs. run index
-- bottom plot is average of that kinematic variable vs. run index
-- for pT, the black line is drawn at the mean value of a guassian fit within the range of the distribution max bin plus-or-minus the RMS

.
.
.