[Core] Describe what contributes to state:modified for each resource type

[dbeatty10]

Link to the page(s) on docs.getdbt.com requiring updates

https://docs.getdbt.com/reference/node-selection/state-comparison-caveats

Tell us more about this update

Folks will often want to understand why a change is or isn't included in state:modified.

It would be great if we documented which things contribute (and possible some examples of which don't).

Reviewers/Stakeholders/SMEs

.

Related GitHub issues

dbt-labs/dbt-bigquery#1192

Additional information

state:modified is determined by check_modified_content which uses the definition of same_contents (which often varies by node type).

For example, a parsed node is considered modified if (and only if), one of the following has changed:

• body
• config
• persisted description
• fqn
• database representation
• contract

Models extend parsed nodes to also include:

• ref representation

But generic data tests use the the following criteria instead:

• config
• fqn

Unit tests use:

• checksum

Macros:

• macro sql

Documentation ("docs blocks"):

• block contents

Sources:

• database_representation
• fqn
• config
• quoting
• freshness
• external

Exposures:

• fqn
• exposure_type
• owner
• maturity
• url
• description
• label
• depends_on
• config

Metric:

• filter
• metadata
• type
• type params
• description
• label
• config

Semantic model:

• model
• description
• defaults
• entities
• dimensions
• measures
• config
• primary entity
• group

Saved query:

• metrics
• group by
• description
• where
• label
• config
• group
• exports

Any other properties not listed above are not taken into account as it relates to state:modified. For example, if a model using the dbt-bigquery adapter defines policy_tags on one of its columns and then changes it subsequently, it will not affect the calculation of state:modified.

[nataliefiann]

I've assigned a medium label to this while we we go through v.1.9 issues but I imagine this will take a day or two to doc