Adding the missing BaseUnits

[lipchev]

Fixes #1463
Fixes #1043

• removed the UnitSystem constructor from the Dimensionless quantities (which was previously throwing)
• As/ToUnit(UnitSystem) for all dimensionless quantities now convert to their BaseUnit (i.e. the "DecimalFraction") *
• As/ToUnit(UnitSystem) for all other quantities refactored using the QuantityInfoExtensions
• added tests for the UnitSystem methods, skipping the tests for all quantities that fail with UnitSystem.SI (with a reason)

There are only two dimensionless quantities (IMO) that don't fit the definition:

• RelativeHumidity: currently has only the Percent unit, I think we should add the DecimalFraction, setting it to be the BaseUnit
• FuelEfficiency: I think this could be defined as "L": -2 with the addition of the MeterPerCubicMeter unit (possibly setting it as its BaseUnit, if we want to satisfy the BaseUnit_HasSIBase test)

You can look for As_UnitSystem_ReturnsValueInDimensionlessUnit if you want to check the rest of the dimensionless quantities.

Regarding the removed BaseUnits (see Force.json or Pressure.json), those were detected by some earlier tests I had in place, regarding the multiplication/division operators where I used the following definition:

• A given operation between two quantities (either multiplication or division) such as A / B = C is only defined if A.Dimensions / B.Dimensions = C.Dimensions
• When the intersection between A.Dimensions and B.Dimensions is the empty set, for every unit of A and B for which the BaseUnits is not Unidefined, and every unit of C, having BaseUnits = A.BaseUnits union B.BaseUnits, it must be true that C.Value = A.Value / B.Value.
• When the intersection between A.Dimensions and B.Dimensions is not empty, and the intersecting BaseUnits of A , B and C are all the same, then again we have the same condition, which I generally refer to as "the conversion coefficient is 1"
• The same logic can be used to infer the unit-conversion coefficient based off the Dimensions and a pair of BaseUnits, but special attention needs to be taken w.r.t. the exponents (e.g. Area is L2 so the unit-conversion coefficients are squares of the ones from Length)
• If we want to extend this definition in the future, we should consider introducing a way to override the "default conversion coefficient" (1)..

Here are some links:
https://en.wikipedia.org/wiki/Dimensional_analysis
https://en.wikipedia.org/wiki/International_System_of_Units#Definition
https://en.wikipedia.org/wiki/Coherence_(units_of_measurement)

[lipchev]

@angularsen I initially thought about not touching the unit definitions, and just [Skip] the tests - but after seeing that the tests failed for 61 quantities, decided that it would be simpler to just add the BaseUnits, rather than add to 61 test files (just to remove the addition in the next PR).

Hopefully you wont find the json files too difficult to review..

PS In #1463 I had totally forgotten that the generator does not (yet) handle the BaseUnits for the prefixes. I wrote some code (long ago) that deals with the issue, but it's a lot, so I decided to leave it out of this PR. Once this is merged, I'll create another PR that deals with that issue alone (which should clear all the [Fact(Skip = "The BaseUnits are not yet supported by the prefix-generator")].

[lipchev]

I took some other notes while preparing this PR:

The initial size for UnitsNet.dll was

2.23 MB (2 339 328 bytes)

1. After adding the extension and refactoring the constructor / As/To(UnitSystem) methods (with 61 quantities failing due to missing SI BaseUnit)

2.20 MB (2 311 680 bytes)

2. After adding all the missing BaseUnits

2.21 MB (2 321 408 bytes)

Still ahead, but if we include the "prefixed" base units- things will likely even out (possibly even setting us back some).

3. After merging from upstream:

2.27 MB (2 384 896 bytes)

PS Finally, here is how this would look with the prefix-generation enabled:

Before merging from upstream

2.23 MB (2 346 496 bytes)

After merging from upstream:

coming soon..

[lipchev]

Also noted the coverage 😄 :

Information
Parser: MultiReport (3x DotCover)
Assemblies: 3
Classes: 293
Files: 298
Line coverage
85%
Covered lines: 30789
Uncovered lines: 5179
Coverable lines: 35968
Total lines: 166435
Line coverage: 85.6%

And now:

Information
Parser: MultiReport (3x DotCover)
Assemblies: 3
Classes: 294
Files: 299
Line coverage
88%
Covered lines: 30364
Uncovered lines: 3765
Coverable lines: 34129
Total lines: 163888
Line coverage: 88.9%

[lipchev]

Oh, forgot about the tests in IQuantityTests - the ones that are testing the UnitSystem methods are now redundant, but we can remove them as we complete the coverage of the As/ToUnit methods..

[angularsen]

Great work! Just a few comments/questions.

[angularsen]

@lipchev As far as I'm concerned, this is ready to go. If you want to make any last minute changes, just merge whenever you want 👍

[angularsen]

Looks good to me, merge whenever you are ready