Less frequent use of back impl

[JNAOB]

CLOSE #301

[jvoigtlaender]

Somehow the implementation below appears too convoluted to me.

We know that the [BinOp] list passed to here will not contain anything aside from Impl, BackImpl, Equi, right?

Then, for example, op == Equi could simply be the first case, and op == Impl || op == BackImpl could be turned into otherwise.

[jvoigtlaender]

This:

Probably in the first few (syntax) tasks we should in the future set (BackImpl, 0) instead of (BackImpl, 1).

could be reflected now by adding code comments to places like this:

logic-tasks/flex/composeFormula.flex

Lines 34 to 47 in d495f1f

	task03 :: ComposeFormulaConfig
	task03 = ComposeFormulaConfig
	{ syntaxTreeConfig = SynTreeConfig
	{ minNodes = 10
	, maxNodes = 14
	, minDepth = 4
	, maxDepth = 6
	, availableAtoms = "ABCDE"
	, minAmountOfUniqueAtoms = 5
	, binOpFrequencies = listToFM
	[ (And, 1)
	, (Or, 1)
	, (Impl, 1)
	, (BackImpl, 1)

i.e., something like

      , (BackImpl, 1)  -- should be (BackImpl, 0) in future

in the last line there.