known spawning/rearing directly from FISS creates confusion regarding barrier cancellation

[smnorris]

In addition to CWF and PSF sources in the known spawning/rearing dataset we now include FISS observations with salmon spawning activity. This adds value, marking "known" spawning locations in bcfishpass outputs.

However, because the model currently cancels out natural barriers downstream of any known spawning/rearing, the result can be confusing in instances of 1-4 poor quality FISS observations upstream of a barrier - only one "known spawning" observation is required to cancel the barrier (vs the expected 5 since 1990).

Consider one of:

• do not cancel barriers just 1-4 of these records, require >5 since 1990
• run some QA to find suspect points and remove

Either option requires a bit of work because the data is materialized in data/user_habitat_classification.csv