feat: add NB_FMD_PURVIEW_LINEAGE_TABLE_COLUMN for standalone column-level Purview lineage

[Copilot]

Adds a dedicated notebook for registering column-level lineage in Microsoft Purview, decoupled from the parallel orchestration notebook. The core addition is dual-mode schema resolution: columns can be read from either a registered Delta table in the attached lakehouse ("table") or any readable path — Delta, Parquet, or CSV — via an explicit ABFSS URI or lakehouse-relative path ("path").

New notebook: NB_FMD_PURVIEW_LINEAGE_TABLE_COLUMN

Schema resolution (source_type / target_type per side):

Mode	Resolution
"table"	spark.read.format("delta").load("Tables/{schema}/{table}")
"path"	spark.read.format(fmt).load(path) — delta / parquet / csv

Purview Atlas entities registered per call:

• Source & target azure_datalake_gen2_resource_set with onelake://{workspace_guid}/{lakehouse_name}/{schema}/{table} qualified names
• Process entity carrying columnMapping (case-insensitive column-name match) enabling column-level lineage in the Purview catalog graph

Exit value — JSON summary (purview_status, mapped_columns, total_runtime) so the notebook is callable from pipelines or orchestration notebooks with inspectable results.

# Table mode — Bronze → Silver
source_type = "table"
source_lakehouse_name = "LH_BRONZE_LAYER"
source_schema = "sales"
source_table = "customer"

target_type = "table"
target_lakehouse_name = "LH_SILVER_LAYER"
target_schema = "sales"
target_table = "customer"

layer = "bronze_to_silver"
purview_account_name = "my-purview-account"

# Path mode — landing file → Bronze
source_type = "path"
source_path = "abfss://workspace@onelake.dfs.fabric.microsoft.com/lh_id/Files/raw/customer.parquet"
source_format = "parquet"