Pretty printing for sparse arrays

[jishnub]

This PR makes the display of block-diagonals similar to that of diagonals, by replacing the zero elements with dots.

After this PR

julia> Diagonal(1:4)
4×4 Diagonal{Int64, UnitRange{Int64}}:
 1  ⋅  ⋅  ⋅
 ⋅  2  ⋅  ⋅
 ⋅  ⋅  3  ⋅
 ⋅  ⋅  ⋅  4

julia> BlockDiagonal([Diagonal(1:2), Diagonal(3:4)])
4×4 BlockDiagonal{Int64, Diagonal{Int64, UnitRange{Int64}}}:
 1  ⋅  ⋅  ⋅
 ⋅  2  ⋅  ⋅
 ⋅  ⋅  3  ⋅
 ⋅  ⋅  ⋅  4

This makes the structure easy to interpret

[codecov]

Codecov Report

Merging #81 (5fd53c3) into master (368887a) will decrease coverage by 0.12%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #81      +/-   ##
==========================================
- Coverage   95.61%   95.48%   -0.13%     
==========================================
  Files           5        5              
  Lines         319      310       -9     
==========================================
- Hits          305      296       -9     
  Misses         14       14              

Impacted Files	Coverage Δ
src/blockdiagonal.jl	85.50% <100.00%> (+0.65%)	⬆️
src/linalg.jl	94.52% <0.00%> (-0.22%)	⬇️
src/base_maths.jl	100.00% <0.00%> (ø)

---

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 368887a...5fd53c3. Read the comment docs.

[mattiasvillani]

As I understand this, this would also automatically cover the case with a block diagonal matrix of banded matrices from BandedMatrices.jl, and similar packages for structured matrices, which would be very nice.