Add terse and is_terse helpers, and use them

[fingolfin]

I deliberately did not yet try to convert any uses of :compact, to make this non-breaking.

This will conflict with PR #1671 but so be it, I can rebase that one.

[lgoettgens]

Complicating matters, there already is a :terse flag in AA, see

AbstractAlgebra.jl/src/PrettyPrinting.jl

Line 564 in 12272bd

terse_level = get(io, :terse, false) ? 1 : 0

and for the effect

AbstractAlgebra.jl/test/PrettyPrinting-test.jl

Lines 268 to 270 in abd59e3

	@test sprint(show, p) == "a + b^2 + c^3"
	@test sprint(show, p, context = :compact => true) == "a + b^2 + c^3"
	@test sprint(show, p, context = :terse => true) == "a+b^2+c^3"

How do we handle this in the context of this new terse naming?

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 86.93%. Comparing base (795382d) to head (4eb3fbf).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1677   +/-   ##
=======================================
  Coverage   86.92%   86.93%           
=======================================
  Files         116      116           
  Lines       29622    29624    +2     
=======================================
+ Hits        25750    25753    +3     
+ Misses       3872     3871    -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fingolfin]

Oh wow, I was not aware of the existing :terse, and the isterse function, at all. Thanks for noticing it.

On the positive side, nothing outside of AA in our stack seems to reference them (there are some occurrences of :terse in Nemo but this s unrelated).

Note that :terse is just used to set the initial value for terse_level (and isterse tests whether the terse level is positive).
But nothing besides a single test seems to set :terse? So is this an unused feature? Or perhaps I missed something using it?

There also is something fishy about the code modifying the `terse_level`: ```julia function set_terse(S::printer) S.terse_level = max(1, S.terse_level + 1) end

function restore_terse(S::printer)
S.terse_level = S.terse_level - 1
end

Note the assymetry: the function increasing the `terse_level` imposes an upper bound of 1, while the function decreasing it does not impose a lower bound... This could conceivable lead to an underrun ? Assuming the two functions are always called in pairs...?
</s>

UPDATE: Nevermind, I just totally misunderstood the code in `set_terse` *sigh*. It sets a *lower* bound of 1, not an upper bound. Ugh.


Anyway, if we drop the use of `:terse` in the IOContext (which seems to be not used by anything), all should be fine?

[fingolfin]

I think PRs #1679 and #1678 take care of the "clash" with names of existing purely internal code.