Support start job from a savepoints dir without specifying savepoint ID

[functicons]

Currently when taking savepoint, users specify a savepoints (parent) dir, Flink automatically generate a savepoint ID and put data in the corresponding subdir. When starting a job from a savepoint, you have to specify the savepoint ID, which means when using the operator, the job args part of the CR has to be changed to point to the specific savepoint.

We want to allow users to only specify savepoints dir in jobSpec, and the operator automatically selects (the latest) savepoint in the dir. This will in turn allow us to update job seamlessly, because savepoint ID becomes transparent.