Harley type contains a coupled inductor. Is this done in your schematic (L7 L8)? You may need to customize it in the simulator options. Test the two windings in both directions relative to one another.
The banana jacks appear to be a means to install C7 C3 in the circuit. One or the other (or both) ought to be hooked up.
The overall loop gain will determine the distortion, so adjusting R to ensure loop gain > 1 but not too high such it saturates and gets distorted. There also may be more than 1 pole with multiple LC so Bode Plots can help choose a narrow band pass filter.
It is important to recognize that all reactive components have a Q at desired frequency.
Depending on size of components, ωL/R =Q is usually < 100 with premium parts higher.
For low f we can ignore self-resonant frequency SRF and Q.
Another Figure of Merit k=R/L which determines cost, current and size.
for 1mH typical values are below;
R = 0.023 Ohms 50 A, $=1000~ R/L = 23 iron core power
R = 10 Ohms 0.2 A , $=2$, R/L = 10k leaded wirewound
R = 110 Ohms 0.046A $=2$, R/L = 110k SMD , Compute Q at operating f to make sure it behaves as expected.
For power inductors in SMPS consider low R/L e.g. = 100 for SMD which are small and inexpensive
For low power Oscillators R/L can be high but Q must be high enough to overcome losses.
For 10 mH similar ratios but must reduce to be Q ~50 or loop gain may be insufficient f. e.g. 120 Ohms/10mH = 12k
--- Updated ---
There are lots of tools to do spectral plots both passive and active.
Here as passive similar filter to the Hartley Osc. The load is from the base. If it rises too much another pole will rise above unity gain.
Then adding realistic DCR resistance of the coil means the gain and Q have dropped.
The bottom line is that one must consider all reactance LC parts have RLC equivalent circuits.
Experience will tell you when it matters to include them in your simulation.
Datasheets will always show ESR for bulk caps and DCR for inductors then SRF, self-resonant freq. to determine Cequiv.
The reason you won't find signal generators with Hartley Oscillators in the audio band is unless it is a high power expensive induction power machine. you need very low R/L values.
If you simulate this design with 20 Ohm DCR in each Inductor, signal will drop 10 fold, harmonics will improve and with inexpensive parts with DCR/L > 100 with won't oscillate for the reasons I mentioned earlier.
Below is with 20 Ohms for each inductor. $>=1000$
When you monitor current in each, you will understand where distortion starts but R/L must be remembered to be realistic.