[REF] WIP : intermediary build_accurate_fields_csv

Refactor to build 3 types of csv in 4 steps :

1. extract **raw CSV** from pdf with camelot

2. build **"accurate" CSV** from raw CSV
    - 1 CSV for each module
    - 1 header for each CSV
    - 1 line for each field
    - No field's cells modification (neither loss of information),
    only mapping which cell is under which CSV's column
    - For each field, adding the field's register name and the
    page number to check back in the pdf
    - option to apply "camelot_row_patches" if necessary

3. build a **"usable" python dictionary** from the "authentic CSV"
    - with "interpreted" values like "required",  "type" and "int_size"

4. build **"usable" CSV** or JSON or whatever from the dict
with **no fields modification**, just built from the dict values

And from these "usabale" dictionnaries :
5. build **"odoo-usable" CSV** _from the dict_ with additional columns
to create Odoo objects from the CSV lines.

Author: clementmbr

.gitignore

@@ -1,4 +1,6 @@
 ecd/
+!ecd/ecd_fields.csv
+!ecd/ecd_registers.csv
 ecf/
 efd_icms_ipi/
 efd_pis_cofins/