Check performance of validation rules

[graste]

As the validation is always run when createEntity is being called on types or when similarity of values is being checked on attribute values its performance is important. Hydrating big entities or a large number of entities should profit a lot on even small gains or performance improvements.

[graste]

The toBoolean() method with it's filter_var method seems to be a low hanging fruit while not losing much functionality.

For ArticleType->createEntity() not using toBoolean() for getOption() is about 7% faster (1000 articles created).

For TextRule apply the speedup of not using toBoolean() for getOption() is about 18% faster on short text and ~5% faster on a 20x longer text. When additionally to not using toBoolean() using the trim() method instead of the UTF8 regex we use atm the speedup is about 35% on short text and about 57% on a 20x longer text (still not that long of a text).

Different combinations were being tested (average of 10x 100000 text rule validations etc.) and were being done in a VM with PHP 5.6. Host system was a i7-2620M CPU @ 2.70GHz.

[graste]

Should try whether trim regex /(*UTF8)^[\pZ\pC]*(?P<trimmed>.*[^\pZ\pC]*)[\pZ\pC]*$/usDS works faster than the one being used at the moment (thanks @ddelbondio).

[graste]

Using preg_replace like $trimmed = preg_replace('/(*UTF8)^[\p{Z}]*+|[\p{Z}]*+$/usDS','',$value); is a bit slower than the default regex with possessive quantifiers (*+; prevents backtracking).

[graste]

Using '/(*UTF8)^[\pZ\pC]*(?P<trimmed>.*[^\pZ\pC]*)[\pZ\pC]*$/usDS' is a 100% faster regardless of * or *+, but breaks tests as not all whitespace is trimmed.