the expected range. Also handle the exception at all call sites.

This is an O(1) implementation compared to the previous O(log n).
In my tests it was faster for all values >4 (int) / >8 (long).

The method that was tested here should have been removed back in https://github.com/h2database/h2database/commit/14664665e4285659b6a48536a490ab0787841470.
Instead it was moved next to its test and was unused ever since.

The method that was tested here was removed back in https://github.com/h2database/h2database/commit/b9ed0d07a8b97fd912d2dc2098f41fe9403fcf11

Merge upstream changes

Update POM file

The corresponding documentation was already removed in
https://github.com/h2database/h2database/commit/197daa58104de6185e08b66e836876eb46b39844

Support drop-columns using brackets (Oracle syntax style) 

too. The type treaded same as CHAR.

This reverts commit e26337e9.

seems to cause various unit test failures.

This reverts commit e26337e9.

seems to cause various unit test failures.

E.g. 
  alter table test drop (a, b) 
works same as  
  alter table test drop column a, b 

fixing incorrect index reuse in AlterTableAddConstraint.java

fixing incorrect index reuse in AlterTableAddConstraint.java ALTER_TABLE_ADD_CONSTRAINT_REFERENTIAL and ALTER_TABLE_ADD_CONSTRAINT_UNIQUE

problem in ALTER_TABLE_ADD_CONSTRAINT_REFERENTIAL was caused by commit 08b98324 by Thomas Mueller <mueller@adobe.com>

he intended to reuse existing index... but it prevents that index from being dropped which can cause problems later
Correct solution would be to rewrite how automatically created indexes (referential/constraint) are managed to create new automatic index when dropping the reused existing index. Until this is solved it's IMHO better to not reuse existing index as it causes database to be effectively corrupted.

Also removing test code Thomas added to TestCases.java

problem in ALTER_TABLE_ADD_CONSTRAINT_UNIQUE was caused by canUseUniqueIndex allowing to use index with more columns than requested as unique index which caused unique constraint to fail as ConstraintUnique checkRow comment says that 'unique index check is enough' and the index is not unique for given constraint if it has more columns