gh-35039: add lazy_string in matrix2.pyx
<!-- ^^^^^
Please provide a concise, informative and self-explanatory title.
Don't put issue numbers in there, do this in the PR body below.
For example, instead of "Fixes #1234" use "Introduce new method to
calculate 1+1"
-->
### 📚 Description
Formatting verbose messages in the method right_kernel could be quite
slow (even when the message is never printed). We use the functionality
lazy_string in order to avoid the formatting.
Also, just calling `verbose()` is slow when just a cache lookup is
needed.
#### Timings
Before:
```
sage: K.<a> = NumberField(x^2 - 2, embedding=1)
sage: A = MatrixSpace(K, 1, 1).one()
sage: %time _ = A.right_kernel()
CPU times: user 23 ms, sys: 3 ms, total: 26 ms
Wall time: 26.1 ms
sage: %timeit A.right_kernel() # cache lookup
2.16 µs ± 70.3 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops
each)
```
After:
```
sage: K.<a> = NumberField(x^2 - 2, embedding=1)
sage: A = MatrixSpace(K, 1, 1).one()
sage: %time _ = A.right_kernel()
CPU times: user 11 ms, sys: 10 µs, total: 11 ms
Wall time: 11.2 ms
sage: %timeit A.right_kernel() # cache lookup
103 ns ± 1.55 ns per loop (mean ± std. dev. of 7 runs, 10,000,000 loops
each)
```
### 📝 Checklist
- [X] I have made sure that the title is self-explanatory and the
description concisely explains the PR.
- [x] I have linked an issue or discussion. **No**, this was discussed
at SageDays 117, we have no linked issue on GitHub for this.
- [x] I have created tests covering the changes. **No**, there are
already tests testing that verbose still works correctly.
- [x] I have updated the documentation accordingly. **No**, there are no
user-facing changes.
URL: https://github.com/sagemath/sage/pull/35039
Reported by: Xavier Caruso
Reviewer(s): Frédéric Chapoton, Sébastien Labbé, Vincent Delecroix
