* testing new nonlinear resolve implementation in file (cat(jets_new_resolve_enable, "/Jets.newresolve.s"))
* LVar returns NULL on expressions without Vars
* numerous bug fixes in the computation of coverings (MM 23.4.2019)

terminology: "combine" replaced by "reduce"

`resolve/nonlin/combine/2` is accepting 2 arguments (to combine two sets, i. e. apply pseudoremainer on cartesian product of two sets) TODO: combinat[cartprod] is very slow  (since )
whereas
`resolve/nonlin/combine/1` apply on single set (pseudoremainder on pairs (but single call on [a,b] ~ [b,a]))

all combining (linear+nonlinear) is done AFTER initial attempt of resolving input eqs

jedeme na vikend

reporting of fails by MM
TODO: BUG FOUND! pokusy2deg3 RESOLVE: <0, 41344> `resolve/data/collect/detect`[0]: No unknowns in res

Sent to MM
# TODO:
# * in remainder fails, references to source nonlinear fails are missing
# * recursive remainder generation is missing (we have single-pass only)
# * `resolve/nonresrat` is not implemented

Jdeme domu.

TODO:
1. `resolve/nonlin/combine/2` must take records as arguments
2. `resolve/data/collect/remainder` must treat correct data, e. g. store parents

`resolve/nonlin/combine` works properly using `resolve/nonlin/combine/2/tool` := `resolve/nonlin/combine/2/rem`: on quadratic polynomials (fixed?)
using linderive() disabled

`resolve/nonlin/combine` works properly using `resolve/nonlin/combine/2/tool` := `resolve/nonlin/combine/2/rem`: on quadratic polynomials (?)