feat: Extend \dt psql command output with shard metadata (#709)

[Adi-Goll]

This PR intercepts the psql \dt command, and add a Shard column to the output.

First a flag to Route that indicates if \dt is being executed so the Shard column is conditionally applied.
A shard_map HashMap is also added to Route that stores table names with their corresponding shard.

This PR also introduces a forward_with_shard() function in backend/pool/connection/binding.rs that
exposes the shard_map property which provides the correct shards be streamed from the query engine.

Finally engine logic was added to populate the new column correctly and handle tables sharded
across multiple databases. The following output uses this config file for testing:

pgdog.toml

[general]
openmetrics_port = 9090
port = 6432
host = "127.0.0.1"
system_catalogs = "sharded"

[[databases]]
name = "mycluster"
database_name = "shard_db_0"
host = "127.0.0.1"
port = 5432
shard = 0

[[databases]]
name = "mycluster"
database_name = "shard_db_1"
host = "127.0.0.1"
port = 5432
shard = 1

[[databases]]
name = "mycluster"
database_name = "shard_db_2"
host = "127.0.0.1"
port = 5432
shard = 2

[[sharded_tables]]
name = "users"
database = "mycluster"
column = "user_id"
data_type = "bigint"

Before this PR:

           List of tables
 Schema |   Name    | Type  | Owner
--------+-----------+-------+--------
 public | only_on_0 | table | ubuntu
 public | only_on_1 | table | ubuntu
 public | only_on_2 | table | ubuntu
 public | users     | table | ubuntu
 public | users     | table | ubuntu
 public | users     | table | ubuntu

Note, the same users table was created on multiple databases, so some deduplication logic was added during streaming, done through an added HashSet property in the QueryEngineContext.
After this PR:

Ex. output:
                List of tables
 Schema |   Name    | Type  | Owner  |  Shard
--------+-----------+-------+--------+---------
 public | only_on_0 | table | ubuntu | 0
 public | only_on_1 | table | ubuntu | 1
 public | only_on_2 | table | ubuntu | 2
 public | users     | table | ubuntu | 0, 1, 2

[Adi-Goll]

I was thinking, maybe it would be better to have a more unique key for the HashMap, maybe instead of just the table name I should use the schema + table name? Thoughts?

[Adi-Goll]

Not sure if it's reasonable to add this for \dt because the original command doesn't do any aggregating or deduplication handling