What are some real world data (e.g., from census data, membership data, visitors of conlang related events, other estimates) on active speakers of constructed languages?
For any particular language maybe 1-2 to dozens. The best way to find out would be to collect as many writing samples as possible that have been posted to the internet. Pick some arbitrary cut off for what counts as "speaking" the language for length. When I did this for toki pona, there were maybe a few hundred people who could write at least a few paragraphs.
Na'vi and Klingon probably have at least a handful to few dozen each. I don't know of any corpus gathering projects for those languages.
The number of people who have purchases learning materials, such as the Klingon dictionary, is thousands of times higher than the number of people who invested the time to learn the language. I look forward so seeing more stats out of Duolingo for the Game of Thrones Languages and Klingon.
As for the acid test of fluency- natives who learned the language from their parents, maybe one or two. (Excluding Esperanto natives- there are a lot of those)