harmonicjada never quite worked and is detached from recent changes in subspacejada (concerning preconditioning etc.). For some interior and non-Hermitian problems it may be useful to have, but right now we don't have applications anyway.
get rid of harmonicjada. Long agao I copied the entire subspacejada routine and made an attempt to use harmonic Ritz values instead, but due to lack of time I never pushed through so that it actually worked. We may consider introducing this option at a later point but should do so without all the code duplication, I'm sure there's a more elegant way.
Fixes #210.
→ <<cset 0836f6daca26>>
