[policy operators] `subset_of` and `essential` example table wrong

[zachmann]

Section 6.1.3.1.8 "Notes on Operators" contains the following tables about combining essential and subset_of, input and what should be the output.

essential	subset_of	(metadata) input	output
true	[a,b,c]	[a,e]	[a]
false	[a,b,c]	[a,e]	[a]
true	[a,b,c]	[d,e]	error
false	[a,b,c]	[d,e]	no parameter
true	[a,b,c]	no parameter	error
false	[a,b,c]	no parameter	no parameter

While this all seems logical, it is not in line with the specification.
The problem is with the two middle lines where subset_of and input are distinct.
According to the subset_of operator description this will result in an empty array, not no parameter. This will be the case for both lines, i.e. also when essential=true there won't be an error, since the metadata parameter has a value set (the empty array) and essential=true will only fail if there is no value set.

In the metadata policy test vectors such a case is included in e.g. n=1562
(The test vector does it according to the spec test, not according to the table).

TL;DR the table is wrong and should read as (or the operators need to be updated):

essential	subset_of	(metadata) input	output
true	[a,b,c]	[a,e]	[a]
false	[a,b,c]	[a,e]	[a]
true	[a,b,c]	[d,e]	[]
false	[a,b,c]	[d,e]	[]
true	[a,b,c]	no parameter	error
false	[a,b,c]	no parameter	no parameter

[selfissued]

@vdzhuvinov can you please comment on this?

[vdzhuvinov]

@zachmann This is correct, the output metadata for these two rows must be []. When I submitted the PR to address issue #182 I did not update the examples in the table. I'll submit a new PR to fix them. Thanks and happy Easter!

[vdzhuvinov]

Here is the fix: #201

[selfissued]

Fixed by #201