fix: freeze data buffer in shard

[evenyag]

I hereby agree to the terms of the GreptimeDB CLA.

Refer to a related PR or issue link (optional)

• Memtable scan performance regression while time series amount < 100K #3467
• Tracking issue for the memtable improvement #2804

What's changed and what's your intention?

This PR fixes an issue that a shard never freezes its data buffer.

• It adds data_freeze_threshold to the Shard and checks whether the shard needs to freeze in write_with_pk_id().
• It enlarges default data_freeze_threshold to 131072

It also corrects some metrics in the PartitionContext as we share it across partitions.

We might need to test the write performance again after merging this PR.

Checklist

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.
• This PR does not require documentation updates.

[fengjiachun]

LGTM

[v0y4g3r]

LGTM