server: reduce the memory usage used by `getAllHistoryDDL`

[bb7133]

Enhancement

TiDB provided a restful API to query all DDL jobs in the history: http://{TiDBIP}:10080/ddl/history.

The initial intention of the API is to debug TiDB in an easy way. However, it has been used by some old ecosystem tools like 'drainer'.

When there are a lot of DDL jobs for a long-running cluster, DDL jobs list will be long and it takes a large amount of memory when querying http://{TiDBIP}:10080/ddl/history.

For the compatibility with old tools, it is reasonable to try reducing the memory usage of getAllHistoryDDL.

[hawkingrei]

I think the parameter that supports the paging is reducing the memory usage.

for example

http://{TiDBIP}:10080/ddl/history?page=1&num=10