Accuracy of LCR meter?

[Martin72]

Ergo it’s better not to worry about a couple of percent and a few counts and work out what the tolerance and device spread of your design is and make sure you window values within that.

Exactly.
You won´t design a coil into a circuit with a tolerance under 5% in most of the cases.
The AL value of the used core makes it impossible for mass production without cost intensive pre-selections.
So tolerances about 20% are not uncommon - Every LCR should handle this..
Of course there are some exceptions, e.g. coils for filters to suppress unwanted harmonics.
We (company I work for) make most of our used coils and transformers ourselves.
And test/adjust them.
We got several different LCRs and therefore different values of the same coil. 
For the most the measuring with one of the LCRs are enough, for other we use resonant test circuits* for adjusting them.
Further we use Voltage-Time-Area function** or, with our self build tester, testing coils with large currents up to 200A***..
Based on this experiences, I can recommend LCRs building like the DE 5000 for most of the needs of a hobbyist.

Martin

*) Example: Filter-Coil, Air-Gap adjustment to reach given value 90µH +/- 3µH.
Using resonant circuit with additonal cap of 10µF/1%, adjust amplifier to 10Vrms, adjust airgap in such manner, that on 1.64....1.65khz the flowing current gets his minimum value.

**) Special case, self satured coils for zero switching, define windings on the coils per feeding in an impulse with high voltage, use math from scope, building integral function for voltage-time-area.

***) High power coils for output filters....normal LCRs didn´t have current, only a few mA max, our self build tester can deliver pulse-current up to 200A, adjustable.
The circuit based on a µC, some electrolytic caps for the current and a IGBT for switching.
L will be calculated based on the formular L= U*t/I (linear current is a must of course)