I can't add anything to the other fine comments but suggest an avenue for experimentation.
You can build a heterodyne oscillator using two crystals which can provide much of the stability of a VXO with a bit of the range of a VFO.
For example a fixed frequency 28060KHz crystal can be "mixed" with a 21060KHz crystal VXO to yield a tuning range of approx 7000 to 7040 KHz.
This is one of my experimental heterodyne oscillators. I built it for a tranceiver LO. I used a 50 ohm feedback amp after the filter followed by a 7 pole low pass filter.
I experimented with taking the output from the gate of the 28MHz xtal and from the source. These were typically cleaner outputs as the drain is rich in harmonics. Two cascode connected J310's can be used for a DGM.
The same concept can be used with simple passive double balanced diode mixers that from experience require less output filtering. Using two high frequency xtals to create a lower frequency puts the xtal harmonics way above the desired one.
Good news is that it can be very stable compared to a L/C VFO. The bad news is that it can be a fairly high parts count and require some form of measurement at the final tune up stage. I could tune the VXO approx 40 KHz with an output that varied only about 2 dB over the tuning range.
I got my crystals from this place:
The worst you get for experimenting is two QRP crystals.
I hope you find a circuit that satisfies your needs. Oscillators of any type are really amazing. It is usually my favorite part of a project. Making a handful of parts oscillate at 7,000,000 cycles per second is still exciting. I have total respect for the UHF and VHF homebrewers.