Improve the conversions benchmarks

[tcojean]

This PR adds some performance and result quality improvement to the
conversions benchmarks.

• Move the construction of matrix_to in convert_matrix function. This allows
  to catch the related exceptions in a way similar to the other benchmarks.
• Do not construct the matrix_to object until the use of copy_from in the
  warmup. This allows to catch non-existing conversions quickly (e.g.
  ELL->COO), without the overhead of loading big matrix data first.
• Catch the AllocationError related exceptions for matrix_from and add a
  completed: false entry to the results on failure to instantiate a matrix
  from the data. This happens a lot when converting from ELL for example,
  which can take a lot of space.

[hartwiganzt]

lgtm

[codecov]

Codecov Report

Merging #317 into develop will decrease coverage by 0.04%.
The diff coverage is n/a.

@@             Coverage Diff             @@
##           develop     #317      +/-   ##
===========================================
- Coverage    98.21%   98.17%   -0.05%     
===========================================
  Files          224      223       -1     
  Lines        17321    17180     -141     
===========================================
- Hits         17012    16866     -146     
- Misses         309      314       +5

Impacted Files	Coverage Δ
omp/factorization/par_ilu_kernels.cpp	0% <0%> (-100%)	⬇️
core/base/mtx_io.cpp	92.85% <0%> (-0.65%)	⬇️
omp/test/factorization/par_ilu_kernels.cpp

---

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 c4f567e...02f393b. Read the comment docs.

[thoasm]

LGTM!