are the files already named, and the new ones you'll record will match them?
If so, I'd suggest adding a sorting prefix before importing them into protools. something that if you sort alphabetically they'll all line up into the way you plan to record the new lines.
You may want to break slightly away from the "single track" thing if your workflow will allow it as well. Maybe do a track per major character or one per level. It'll help with organization.
Once your ref files are properly named, you can import them into PT, sort the bin alphabetically, set the drag property to "left to right", and drag them into a track.
From there I'd strongly suggest investing in quickeys, as the scripting functions there will help tremendously when dealing with thousands of files in a single session.
One trick that quickeys can do would be to automate a loop that'll add x amount of time between each region - say 5 seconds. Just copy 5 seconds of silence, put PT in shuffle mode, then create a quickeys loop that goes: pause 0.5 sec, paste, tab, tab.
Now you have every file on your timeline, properly named and with 5 seconds between each line. If you end up needing to find any given line on your timeline then command+f, type the name, click the file in the region bin, and hit the left arrow key to center up the region you selected in front of you.
of course, this is making a ton of assumptions about your workflow but given the information you provided that's where I'd start.