[tests][oceanbase] Fix oceanbase test failure, possibly caused by some interactions among cases

[yuxiqian]

This fixes a mysterious test failure in OceanBaseMySQLModeITCase like this. Recently it is happening quite frequently, possibly brought in by #3439.

This could be stably reproduced locally, by running OceanBaseMySqlModeITCase locally, and make sure testMetadataColumns is executed right after testSnapshotOnly.

This test case will fail, because instead of getting expected

+I(test,inventory,products,101,scooter,Small 2-wheel scooter,3.1400000000), 
+I(test,inventory,products,102,car battery,12V car battery,8.1000000000), 
+I(test,inventory,products,103,12-pack drill bits,12-pack of drill bits with sizes ranging from #40 to #3,0.8000000000), 
+I(test,inventory,products,104,hammer,12oz carpenter's hammer,0.7500000000), 
+I(test,inventory,products,105,hammer,14oz carpenter's hammer,0.8750000000), 
+I(test,inventory,products,106,hammer,16oz carpenter's hammer,1.0000000000), 
+I(test,inventory,products,107,rocks,box of assorted rocks,5.3000000000), 
+I(test,inventory,products,108,jacket,water resistent black wind breaker,0.1000000000), 
+I(test,inventory,products,109,spare tire,24 inch spare tire,22.2000000000), 
+U(test,inventory,products,106,hammer,18oz carpenter hammer,1.0000000000)

values, it is actually receiving:

+I(test,inventory,products,101,scooter,Small 2-wheel scooter,3.1400000000),
+I(test,inventory,products,102,car battery,12V car battery,8.1000000000), 
+I(test,inventory,products,103,12-pack drill bits,12-pack of drill bits with sizes ranging from #40 to #3,0.8000000000), 
+I(test,inventory,products,104,hammer,12oz carpenter's hammer,0.7500000000), 
+I(test,inventory,products,105,hammer,14oz carpenter's hammer,0.8750000000), 
+I(test,inventory,products,106,hammer,16oz carpenter's hammer,1.0000000000), 
+I(test,inventory,products,107,rocks,box of assorted rocks,5.3000000000), 
+I(test,inventory,products,108,jacket,water resistent black wind breaker,0.1000000000), 
+I(test,inventory,products,109,spare tire,24 inch spare tire,22.2000000000), 
+I(test,inventory,products,101,scooter,Small 2-wheel scooter,3.1400000000), 
+I(test,inventory,products,102,car battery,12V car battery,8.1000000000), 
+I(test,inventory,products,103,12-pack drill bits,12-pack of drill bits with sizes ranging from #40 to #3,0.8000000000), 
+I(test,inventory,products,104,hammer,12oz carpenter's hammer,0.7500000000), 
+I(test,inventory,products,105,hammer,14oz carpenter's hammer,0.8750000000), 
+I(test,inventory,products,106,hammer,16oz carpenter's hammer,1.0000000000), 
+I(test,inventory,products,107,rocks,box of assorted rocks,5.3000000000), 
+I(test,inventory,products,108,jacket,water resistent black wind breaker,0.1000000000), 
+I(test,inventory,products,109,spare tire,24 inch spare tire,22.2000000000), 
+I(test,inventory,products,101,scooter,Small 2-wheel scooter,3.1400000000), 
+I(test,inventory,products,102,car battery,12V car battery,8.1000000000)

which makes totally no sense, just like previous data records have not been cleaned up correctly. Running either testSnapshotOnly or testMetadataColumns independently is OK, so this PR creates & destroys OceanBase containers for each test case to avoid breaking CI, as a temporary hackaround.

I'm a little worried that this behavior might not be caused by a bad test case but reflects some connector implementation issue. Needs some advice from @whhe.

[yuxiqian]

Another typical failure case is when testAllDataTypes case is running, waitForSinkSize never collects enough rows, and keeps running for an hour, since there's no any timeout mechanisms.

[whhe]

To avoid reading change events generated by another test case, how about using UniqueDatabase? Starting a new container for each test case may significantly increase the testing time.

There should be a timeout for waitForSinkSize method. Please help to fix it here.

[yuxiqian]

Done, please check.

[yuxiqian]

Rebased with master, could @ruanhang1993 please take a look?

[ruanhang1993]

LGTM