[Improvement](statistics)Improve show analyze performance.

[Jibing-Li]

The origin show analyze code would traverse the task id to task map for each job to get all the tasks for this job, which is very slow.
This pr keep the task ids for each job in the job info class, so that we can get the tasks by task ids directly.
To show 1100+ jobs with 11000+ tasks, the time costing is reduced from 0.14s to 0.01s.

Further comments

If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...

[Jibing-Li]

run buildall

[Jibing-Li]

run buildall

[hello-stephen]

(From new machine)TeamCity pipeline, clickbench performance test result:
the sum of best hot time: 45.31 seconds
stream load tsv: 512 seconds loaded 74807831229 Bytes, about 139 MB/s
stream load json: 20 seconds loaded 2358488459 Bytes, about 112 MB/s
stream load orc: 65 seconds loaded 1101869774 Bytes, about 16 MB/s
stream load parquet: 31 seconds loaded 861443392 Bytes, about 26 MB/s
insert into select: 29.5 seconds inserted 10000000 Rows, about 338K ops/s
storage size: 17162103814 Bytes

[morningman]

LGTM

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

PR approved by anyone and no changes requested.

[Kikyou1997]

LGTM