This patch fixes a bug/inconsistency where permutation vector was not

set when there were zero columns in LU factorization.

This release includes major speedups for the acb_theta module, using

strides instead of row pointers in all matrix types, and a lot of
additions to the gr module, the Generic Rings module in FLINT.

For a complete changeset, see https://flintlib.org/doc/history.html.

This patches fixes:

* Installation of header for out-of-tree builds (#2263)

* rbit instruction was incorrectly assumed to exist on Aarch64 (#2264)

* Assertion failure because of an unitialized variable (#2273)

This patch fixes:

* Recognition of AMD Zen during configuration

This release includes the new modules `mpn_mod` and `nfloat` for packed

fixed-size modulo arithmetic and packed fixed-size floating-point
arithmetic, respectively.  Moreover, it includes major work on the `gr`
modules.

For a complete changeset, see https://flintlib.org/doc/history.html.

Second release candidate of FLINT v3.2.0

This includes the new modules `mpn_mod` and `nfloat` for packed
fixed-size modulo arithmetic and packed fixed-size floating-point
arithmetic, respectively.  Moreover, it includes major work on the `gr`
modules.

For a complete changeset, see https://flintlib.org/doc/history.html.

First release candidate of FLINT v3.2.0

This includes the new modules `mpn_mod` and `nfloat` for packed
fixed-size modulo arithmetic and packed fixed-size floating-point
arithmetic, respectively.  Moreover, it includes major work on the `gr`
modules.

For a complete changeset, see https://flintlib.org/doc/history.html.

Merge pull request #1999 from edgarcosta/release-version-3.1

3.1.3-p1

Fix bugs in the `fq_default` context initialisers

Some explanations for mpn_ctx_mpn_mul