[3.x] Remove Type::closureToPhp() optimization

[GromNaN]

Q	A
Type	improvement
BC Break	yes
Fixed issues	#2870

Remove the Type::closureToPhp() that is used for generating the hydrator classes.

Doing some benchmarks, I don't see a significant difference.

Before

./vendor/bin/phpbench run benchmark/Document/HydrateDocumentBench.php
PHPBench (1.4.3) running benchmarks...
with configuration file: phpbench.json
with PHP version 8.5.0, xdebug ✔, opcache ❌

\Doctrine\ODM\MongoDB\Benchmark\Document\HydrateDocumentBench

    benchHydrateDocument....................I9 - Mo1.221ms (±11.53%)
    benchHydrateDocumentWithEmbedOne........I9 - Mo1.117ms (±5.89%)
    benchHydrateDocumentWithEmbedMany.......I9 - Mo1.225ms (±7.01%)
    benchHydrateDocumentWithReferenceOne....I9 - Mo1.214ms (±4.15%)
    benchHydrateDocumentWithReferenceMany...I9 - Mo1.199ms (±9.18%)

Subjects: 5, Assertions: 0, Failures: 0, Errors: 0

After

./vendor/bin/phpbench run benchmark/Document/HydrateDocumentBench.php
PHPBench (1.4.3) running benchmarks...
with configuration file: phpbench.json
with PHP version 8.5.0, xdebug ✔, opcache ❌

\Doctrine\ODM\MongoDB\Benchmark\Document\HydrateDocumentBench

    benchHydrateDocument....................I9 - Mo1.241ms (±5.25%)
    benchHydrateDocumentWithEmbedOne........I9 - Mo1.161ms (±6.17%)
    benchHydrateDocumentWithEmbedMany.......I9 - Mo1.222ms (±7.48%)
    benchHydrateDocumentWithReferenceOne....I9 - Mo1.238ms (±7.36%)
    benchHydrateDocumentWithReferenceMany...I9 - Mo1.215ms (±4.35%)

Subjects: 5, Assertions: 0, Failures: 0, Errors: 0

[malarzm]

It is 10% slower though

[GromNaN]

It is 10% slower though

That's true, but benchmarks vary quite a bit. In any case, we know that it cannot be more performant.