After 5 years of better computers and some development on PT, the slowdown may not be as bad as it used to be.
But in general extremely long windows in PT slow it down. For example, if you opened the entire book of Exodus in a single view and started scrolling through, it might start lagging. That is particularly true for those of us using complex fonts which take extra computing power--it may be less noticeable if you're using a Latin script. This is part of the reason people almost always edit using "View-->Show all chapter" in its un-checked state.
With the glossary, that becomes a problem because the PT developers didn't create a method for breaking it up into chapters (and thus more manageable chunks). So by default turning off "View-->Show all chapters" doesn't help.
What I would suggest would be to start work on a glossary. Then, if you start to get slowdown, artificially split your glossary into chapters. Doing this will mean the linking with Biblical Terms stops working, so only use the chapter option if absolutely necessary.
Note that the slowdown only happens if that Glossary book happens to be open. It doesn't slow PT down across the board at all times.
-----------------
I just tested and, yes, the issue with flagged notes in the glossary is still a problem. I would call it a fairly major bug, personally.
When PT creates a flag, it saves some "context" of the surrounding words in order to place the flag in the right place. In regular books, that context is the verse. But since the glossary isn't broken into verses, it saves the entire chapter (which may be the entire glossary) each and every time you create a flag.
For example, our Glossary file is currently about 300kb long. That means that every time you create a new flag (or reply to a flag?), your Notes.xml file gets 300kb longer. You can see how that would add up quickly. (Breaking the glossary into chapters does make this problem less severe.)
In actual practice this doesn't cause us any problems. The team is likely only adding a few flags at a time, so each send/receive only uses one or two MB of data. But if we ever need to do a fresh S/R on a brand new project, it can take a while. One of the projects I'm involved with has 400+mb just of notes.