Are those old glass capacitors actually worth using, or just pretty?

Those are almost certainly fine to use. Glass dielectric parts from Corning are about the most robust caps ever made. The glass body doesn't age, doesn't absorb moisture, and doesn't drift the way ceramics or electrolytics can. The only weak points are the end seals where the leads enter, so look for hairline cracks there or any green crud on the lead wires. If the seals look clean and intact, the guts are basically as good as the day they were made. With just a basic multimeter you can do a decent sanity check. Put it on the ohms range and touch the leads to the cap. You'll see the reading climb as the meter charges the cap through its internal circuit, then it should settle at open or over-limit. Anything that settles at a low resistance reading is leaky and you toss it. If your meter has a capacitance setting, that's a bonus, but a lot of cheap meters struggle to read accurately down at 4700pF, so don't panic if the number looks off by a bit on that range.

As for whether to use them in the rebuild, I'd say use them. A faithful rebuild is the whole point of the project, and an oscillator is exactly the kind of circuit where those parts shine. They have virtually no dielectric absorption, essentially no microphonics, and rock solid temprature behavior. A modern C0G will match them on paper, sure, but it won't match them on sentiment, and the electrical difference in that circuit will be nothing you can measure.

My approach with stuff like this is to use the original parts where they test good and stash a couple in the box as keepsakes. That way the build works and you still have a piece of your uncle's bench sitting in the drawer. Win on both counts.

Ha, those take me back. We used to pull the odd glass cap out of mil-spec surplus boards back in the 70s and they were the first thing the old hands grabbed for anything timing related. Never once saw one go leaky, though I did crack one soldering the leads too close to the body. Heat them with longnose pliers as a heatsink and clip your leads before they hit the solder iron. One more thougt: if the oscillator has a trimmer in there, leave a little adjustment range either way, because old glass sometimes runs a touch under nominal compared to what the modern parts measure.

I'd say build with them, yeah. I put my granddad's old parts back into a radio he owned and it just feels right when it's running. One tip on the soldering though, heat travels up those leads fast into the glass, so grip the lead with pliers between the cap and the joint to soak up some of it. Cracked one that way myself and it was annoying to find. Also worth writing down which caps came from where, future you will thank you for it.

One thing I'd add is check the actual value before assuming the marking is still what it says. I measured a batch of old caps from a similar estate box and a couple of the 472 parts had drifted up around 5 percent, not leaky at all, just off enough to matter if the oscillator frequency is critical. A cheap meter on its capacitance range is close enough for comparing them against each other even if the absolute number is a bit optimistic. Also worth measuring them in circuit position one at a time so you know which stray readings come from the wiring.

Update: tried the pliers trick on the leads while soldering up a test setup and it made a huge difference, no cracks so far. What's still bugging me is the drift point someone raised, since my meter has no capacitance function and one of these coud be off without me knowing. Does a cheap handheld cap tester give a trustworthy reading on old glass parts, or do they read wierd?