In FrontendService, we may create some connection context and set it as a thread local varaible.
These context should be removed from thread local after call.
Otherwise, it may be reused by other thread incorrectly.
1. Add `BINARY` keywords
`BINARY` is used for tell MySQL to treat a String(or String column) as case sensitive.
eg: `"abc" = "ABC"` is true, but `BINARY "abc" = "ABC"` is false.
But in Doris, `"abc" = "ABC"` is false by default.
I add this `BINARY` keyword just for compatibility, it will take no effect.
2. Add `PARTITIONS` and `AUTO_INCREMENT` as reserved words.
`PARTITIONS` is the table name in `information_schema` database.
`AUTO_INCREMENT` is a column name of a table in `information_schema`
- `CANCEL MATERIALIZED VIEW TASK taskId on mvName`
- CANCEL MATERIALIZED VIEW TASK, tasks("type"="mv") and jobs("type"="mv") support check auth use priv of mv
- tasks and jobs add column mvName and mvDbName，you can use `select * from tasks("type"="mv") where MvName="xxx"` get all tasks of mv
- fix `desc mv all` error
- fix p0 The task sequence is incorrect
* Revert "[fix](move-memtable) only check missing tablets when commit info is not empty (#29326)"
This reverts commit 4634c723d7fb86f434060c0e6fa6168d92630cb9.
* Revert "[fix](move-memtable) check missing tablets before commit (#29223)"
This reverts commit fb0ed8c253f5241c23c71fb88f031604b0bf5a1f.
Some sink may not get any success tablets, because load stream will only report them at the last CLOSE_LOAD.
For these sinks (no commit info), we should not check missing tablets since it's expected to be empty.
Both tables simply need to change the tablet id to get the same result from the show proc statement.
before：
image
Just change the tablet id to get the result, 10139 does not belong to the tablet in the table id of 10127
later:
image
Additional deleted filter condition will be materialized column be at the end of the block,
after _output_column_by_sel_idx will be erase, we not need to filter it,
so erase it from _columns_to_filter in the first next_batch.
Eg:
delete from table where a = 10;
select b from table;
a column only effective in segment iterator, the block from query engine only contain the b column,
so no need to filter a column by expr.
