Turn on reduction for noodlification

[jurajsic]

This PR fixes a bug where noodlification did not reduce (nor trim) automata in resulting nooodles.

Needs to be benchmarked, it is possible that simulation might slow it down, or that we should just turn on trimming.

[vhavlena]

Maybe the simulation reduction will be too expensive. We could add option "trim" to do just the trimming. But we will see after the results.

[vhavlena]

I suggest first to finish and merge the PR #104 and then try the effect of the noodle reduction.

[vhavlena]

The results (z3-noodler-feada45-a57f582 is the PR version, z3-noodler-1482571-a57f582 is the pyex-opt version -- NOT devel)

# of formulae: 82972
###################################################################################
####                                   Table 1                                 ####
###################################################################################
| method                     |    max |     mean |   median |   std. dev |   TO+MO+ERR |   unknowns |
|----------------------------|--------|----------|----------|------------|-------------|------------|
| z3-noodler-feada45-a57f582 | 116.1  | 0.494545 |     0.02 |    3.67665 |        2198 |       2011 |
| cvc5-1.0.8                 | 119.36 | 0.515113 |     0.02 |    5.2351  |        1723 |          2 |
| z3-4.12.2                  | 119.29 | 1.49018  |     0.04 |    8.17615 |        2623 |        123 |
| z3-noodler-1482571-a57f582 | 115.42 | 0.463528 |     0.02 |    3.42857 |        2465 |       2098 |

• the version z3-noodler-1482571-a57f582 lacks some optimisations currently in devel (hard equations, replace_re, lex ordering), the improvement of this PR is not caused only by the noodle reduction. Ignore woorpje.
• it seems there is some positive effect (e.g., on kaluza we have 80 TOs less now)

[vhavlena]

Should I merge the PR?

[Adda0]

I would merge the PR.