Fix exception in datacube dataset update for numeric keys

[jeremyh]

The dataset update command threw an exception when one of the unsafe changes was a numeric key:

File "/g/data/v10/public/modules/dea/20200526/lib/python3.6/site-packages/datacube/index/_datasets.py", line 284, in <genexpr>
    full_message = "Unsafe changes at " + ", ".join(".".join(offset) for offset, _, _ in unsafe_changes)
TypeError: sequence item 2: expected str instance, int found

This fixes that, and updates the product update, metadata update and dataset update commands to display document changes more consistently with each-other.

Other minor issues fixed:

• It now tells you which document the error occurred in (since we often have multiple documents in one yaml)
• The unsafe value changes are now shown as warnings so that they display by default without needing to rerun the command with higher verbosity.

Issues not yet fixed:

It's a pain to make changes to a multi-document yaml. I wish I could tell it "update the ODC to match my whole yaml", without having to pick apart which inner-documents need updating and which need adding. Are we sure add and update need to be mutually exclusive?

(An example of multi-doc yamls are DEA's metadata-types and products)

[codecov]

Codecov Report

Merging #983 into develop will increase coverage by 0.02%.
The diff coverage is 99.43%.

@@             Coverage Diff             @@
##           develop     #983      +/-   ##
===========================================
+ Coverage    92.95%   92.98%   +0.02%     
===========================================
  Files           97       97              
  Lines         9649     9703      +54     
===========================================
+ Hits          8969     9022      +53     
- Misses         680      681       +1     

Impacted Files	Coverage Δ
datacube/index/eo3.py	100.00% <ø> (ø)
datacube/ui/click.py	84.31% <0.00%> (-0.11%)	⬇️
datacube/utils/geometry/__init__.py	100.00% <ø> (ø)
datacube/api/core.py	98.37% <100.00%> (+<0.01%)	⬆️
datacube/api/grid_workflow.py	92.91% <100.00%> (-0.22%)	⬇️
datacube/drivers/netcdf/_write.py	100.00% <100.00%> (ø)
datacube/drivers/netcdf/driver.py	100.00% <100.00%> (ø)
datacube/drivers/postgres/_core.py	95.60% <100.00%> (ø)
datacube/index/_datasets.py	94.60% <100.00%> (+0.25%)	⬆️
datacube/index/_metadata_types.py	94.79% <100.00%> (-0.06%)	⬇️
... and 16 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update dae57fb...8691ebc. Read the comment docs.

[omad]

Thanks for the thorough changes Jeremy, looks good to me.

Could you please make an issue to discuss the semantics of add vs update.

My thoughts in brief (I'll copy to an issue when it exists):

• Multi-document YAMLs add significant complexity to quite a few of our operations
• They are useful for storing related resources together,
• When editing it's not obvious if or how many extra documents are in the same file.
• Kubernetes also uses YAML for storing configuration, and allows multiple completely different records in the same file
• It's CLI apply command takes your suggested approach of "make the database match what's in my file", regardless of whether it's adding/update, while also checking that it's a safe change.
• If we're going to keep supporting multiple document YAMLs, I think we should support this operation. But I'm not sure what name to use, and I'm not sure if we should keep add and update.

Crazy Damien idea that should not be seen

Let's implement an ODC Kubernetes Operator, which you interact with to update an ODC Database via Kubernetes tools