fix(nyc-taxi): match instance name exactly to avoid prefix collision

[yanou16]

TL;DR

The instance filter uses a substring test, which collides when one instance
name is a prefix of another ("nyc_taxi" ⊂ "nyc_taxi_pipeline"). Fixed by
matching the exact instance name.

Problem

The nyc-taxi helper scripts filter discovered dataset URNs by instance using a
plain substring test:

if platform_instance not in urn:      # add_lineage.py
if platform_instance not in entity.get("urn", ""):   # add_metadata.py

This is wrong when one instance name is a prefix of another. The dataset
ships two instances, nyc_taxi and nyc_taxi_pipeline, and "nyc_taxi" is a
substring of "nyc_taxi_pipeline". So when running for --instance=nyc_taxi,
the filter also lets nyc_taxi_pipeline URNs through. Same-named tables
(staging_trips, mart_daily_summary, ...) then overwrite each other in
urn_map, and lineage/metadata for the clean nyc_taxi instance gets emitted
onto nyc_taxi_pipeline URNs — leaving the nyc_taxi instance with no lineage.

Fix

Match the exact instance by checking for ",{instance}.". Inside a dataset URN
the instance name is delimited by a comma and a dot
(...,nyc_taxi.main.staging_trips,...), so ",nyc_taxi." does not match
",nyc_taxi_pipeline.". Applied in both add_lineage.py and add_metadata.py.

Impact

• datahub ... add_lineage.py --instance=nyc_taxi / --all now writes lineage
  to the correct instance.
• No behavior change for single-instance users or for nyc_taxi_pipeline.

How to reproduce (before the fix)

python add_lineage.py --all
# then query lineage for nyc_taxi.staging_trips -> 0 downstream (should be 4)

[lakshay-nasa]

Thanks @yanou16! Clean fix, applied consistently across both scripts 🙌