First, as more of a philosophical angle, it's better to look for high value client, not high value projects.
[Quick note, I've not really used UpWork since the departure from oDesk, and I'm not at all a fan of the upcharges just to get better access to data.]
First, your specific questions:
If a project has too many proposals already, should I not bother? How many is "too many"?
When I was bidding on projects, my rate would always be near the top (if not the highest). I wasn't concerned with how many proposals there were, what I looked at was the average rate vrs the average interview rate. If the interview rate was at least a little higher, it told me the client's primary interest was not the lowest hourly rate.
It now looks like UpWork charges for access to those kind of stats.
If the project already has "hires", should I not bother?
This depends on the type of work you're doing. If I had to choose between two jobs, and one already had a hire, I'd probably pick the other. But this could also be a useful indicator about the client. If it's a job where more than a single contractor are obviously needed, can you identify the hire? If so, what does that tell you about the client?
If the project is still open after a week, and has no recent activity, should I assume the client has moved on and never gotten around to closing it out?
Time itself isn't necessarily an indicator of that. Have they moved anyone to interview (if so, perhaps they've just not found anyone yet)? What do the average bids tell you (lots of low bids may mean they're waiting for someone with what the consider 'reasonable' pricing)?
Some general tips:
I'd always use bidding on projects to expand my area of expertise. As a contract developer doing API integrations, if there was an API I wanted to play with, or wanted to become an 'expert' in, I'd focus my bidding on those jobs.
Think about your reputation when you bid - specifically, does this job seem like it will end successfully with the client leaving positive feedback.
This may no longer be true, but oDesk used to match you in search results to not only your profile, but the keyworks in the jobs you bid / won. Keep that in mind, because what terms your profile matches will affect the job invites you get.
And on that topic, your profile may be the thing you can improve the most. Right now you're asking how to validate that a job is worth using one of your available bids. Wouldn't it be better to just filter inbound requests? How valuable is a good profile? Valuable enough that mine has been copied multiple times.
(Which was an interesting story, I found out about that interviewing another contractor for a client, and thought some of their profile sounded familiar. Turns out it was mine.)
I touch on some other tip in this rather dated guest post on the oDesk blog, parts are still relevant.