feat: parquet datasource

[adrienaury]

Add a new datasource type supported by lino : parquet

The initial implementation supports :

• no ingress description definition, by default lino pull will follow the parquet schema
• a single table by file, with name "data" by convention
• read only, can't update existing file (lino push will overwrite existing file)

Use the provided example file to test parquet features : test.zip

lino dataconnector

$ lino dc add myfile 'parquet://path/to/test.parquet'
successfully added dataconnector
$ cat dataconnector.yaml
version: v1
dataconnectors:
  - name: myfile
    url: parquet://path/to/test.parquet
    readonly: false

lino table

$ lino table extract myfile
lino finds 1 table
$ cat tables.yaml
version: v1
tables:
  - name: data
    columns:
      - name: col_type_binary_string
      - name: col_type_int_32
      - name: col_type_boolean

lino pull

$ lino pull myfile
{"col_type_binary_string": "6FBFXK4V82XQA4EZ0UCY8PZZN2B256", "col_type_int_32": 324442164, "col_type_boolean": true}