VSD and harmonic mitigation requirement

Where/ how  is the 35% measured ?  If it is the waveform of the mains current I'd expect the DNO to be unimpressed unless that is a small fraction of the total connected load  ! Figures like <5% target, 8% upper bound under worst case are commonly specified at point of connection.

If it is the voltage or the current on the side of the customer's wiring and loads, well its a matter for them, but problems are entirely possible, unless the load is very robust - a large motor that is mostly under-run may be OK or anything producing heat - but then you dont need the VSD.
Depending on its exact composition, 35% THD is pretty badly non-sinusoidal and may be enough to overheat magnetic cores as well as increase losses in the copper.

Mike

Ultimately this is a matter between your customer and their DNO (or National Grid if directly connected).  The DNO will care about what appears at the interface between the site and their network so what else is connected on the site will have an effect, as will the electrical "distance" to other customers and the impedance (e.g. prospective short circuit capability) of the supply.

I suspect that the DNO will use ENA EREC G5 as a starting point but this is a discussion that needs to be had with the customer who in turn should involve the DNO if it has a material effect at their interface.

Indeed,  and if you dont have one, Dcode website has an on-line  a copy ENA_EREC_G5_Issue_5__2020_.pdf

On page 21, is indeed  '5%' listed as a THD planning level for 400V 3 phase ...  lower levels are generally used at higher voltage distro. 
I'd expect an application  for an uncorrected  35% THD load on the network to be laughed off stage. 

Mike.