I'm only speaking of the data you linked. If you have the data-center data, feel free to post it.
Yes, it is a single scan. There are previous scans and you can compare to them (e.g. Population change charts). But the character data you see is still just a snapshot of the lodestone at one point in time. The profile only includes the gear of the current job, because that is all you can check. It even says so right before the weapon ilvl charts (basically, exactly what I'm saying):
"Class at the time of Lodestone character data acquisition
It is the class reflected in Lodestone at the time of data collection, and not necessarily the main job class."
It is no coincidence that you only get weapon ilvl data for one character per profile:
Combat job profiles (sum of all weapons in the ilvl distribution chart) + Crafter/Gatherer profiles + Blue Mage profiles + Class profiles (no soulstone) = Current active profiles
Yes, over a large enough data set, being on an alt makes no meaningful difference. But that is not the kind of data you have here. You have one snapshot.
If you sum up all unlocked jobs and divide by the active profiles, you'll get an average of 9 unlocked jobs per profile. Statistically that means that every healer main accounts for at least 6 non-healers and every tank main accounts for at least 5 non-tanks.
Having unlocked multiple jobs is not the fringe case. Having only unlocked your main role is the fringe case. This is also reflected in the overwhelming majority of DPS jobs at lower levels - meaning that most people have unlocked DPS jobs, but are not leveling them.
If you sum up ilvl440+ weapons you'll get a healer:tank:dps ratio of around 1:1:3, which seems pretty damn healthy in terms of healers to me.
Presumably, this would indicate what players are maining, but for reasons mentioned before, the data does not and can not tell you what people are maining.
Besides, you've made a sweeping statement that "healers are in really low numbers". Why would I only look at ilvl440+ if I'm trying to address that?
I'm not saying that the data is pointless or that it says nothing. All I'm saying is that it does not support this conclusion: "healers are in really low numbers".