After some additional test for #476 and more checking with the source
code I found out that I make a mistake in the grammar. The
MySQL/Postgres and the Derby/Oracle syntax should be different
productions.

Document fetch first row(s) only

Fix doc generation

In order to make the doc generation for #476 working I had to fix some
errors in existing files. These cover

 - lines which are too long
 - incorrect indentation
 - invalid markup
 - missing javadoc

Since 1.4.177 H2 supports FETCH FIRST x ROW ONLY however this is not
reflected in the grammar documentation. This covers all the variations
of the syntax.

There are 3 `help.csv` but I believe I got the right one. Some of the
checked in files seem to have been automatically changed and I don't
know if I should have committed those as well.

Fixes #475

Add alias for RANDOM_UUID

Fixes h2database/h2database#467 Deadlock between Sequence.flush() and…

* Custom data types handler

* Updated doc, changelog, dictionary. Fixed review comment.

* Extended the custom types handler interface

* Fixed test

* Fixed review comments

* Added note on ABI stability of the feature to docs

to make it more obvious

Updated commercial support information

Use System.nanoTime() to check timeouts

System.currentTimeMillis() is influenced by system time corrections
whereas System.nanoTime() is strongly sequential. So it is better to use
System.nanoTime() when checking for timeouts and elapsed time.