unchanged source lost harvest_object_id after each reharvest

[FuhuXia]

Found this issue when researching #4348.

For each record in the source, a reharvest will delete existing harvest_object_id and create a new harvest_object_id in DB, even the record has not changed. But it keeps the old harvest_object_id in the Solr. As a result, DB and Solr is out of sync after each reharvest. Before db-solr-sync run, the dataset page on the UI lost harvest_object_id.

example sources:
/harvest/national-transportation-atlas-database-ntad-metadata
/harvest/edit/fema-r02

How to reproduce

Using anyone of dataset in the source as example.
Locate the harvest_object_id in the API.
Reharvest
Now the dataset on the UI has lost its harvest source info.
reindex the package using ckan command
Now harvest_object_id in the API is updated. Now UI has harvest source info.
Reharvest
Now the dataset on the UI has lost its harvest source info.

Expected behavior

If a WAF record has no changes, gather should not create any fetch job.

Actual behavior

processing unchanged record and harvest_object_id is out of sync.

Sketch

[Notes or a checklist reflecting our understanding of the selected approach]

[FuhuXia]

The issue is due to ckanext-spatial's way of detecting IIS server is out of date. Will create a upstream PR.

update:
Upstream PR created.

[FuhuXia]

fixed verified by reharvesting /harvest/national-transportation-atlas-database-ntad-metadata.