Avoid reshape when materialising DualLayout views of matrices

[dlfivefifty]

sub_materialize(::DualLayout, V) for a view V = view(A, :, jr) calls _dual_adjoint/_dual_transpose, which do view(parent(V)', jr). This assumes parent(V) is an adjoint/transpose of a vector. But DualLayout is also reported by 1×n matrices that wrap one, e.g. BlockedMatrix(v', ...) or MyMatrix(v'). In that case parent(V)' is an n×1 matrix, so view(·, jr) is linear indexing, and Base reshapes. For infinite arrays this throws:

using BlockArrays, InfiniteArrays, FillArrays
d = BlockedArray(BlockedVector(2.0*(0:∞), (blockedrange(Ones{Int}(∞)),))', (Base.OneTo(1), blockedrange(Ones{Int}(∞))))
d[:, Block.(1:3)] # ArgumentError: Cannot multiply 0 * ℵ₀

Here the reshape's ind2sub calls divrem(i, ℵ₀). This came up indexing a BandedBlockBandedMatrix whose data is such a BlockedMatrix.

The fix adds a _dual_view helper: for a vector it's view(v, jr) as before, and for a matrix it's view(A, jr, only(axes(A,2))). The behaviour for genuine adjoint/transpose vectors doesn't change.

The new tests check that _dual_adjoint/_dual_transpose of a wrapped adjoint/transpose (real and complex) don't produce a ReshapedArray. They also cover the infinite case MyMatrix(InfVec()')[:, 1:3]. Both fail without the fix. The version is bumped to 1.16.1.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.81%. Comparing base (4a778aa) to head (1d39d74).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #291      +/-   ##
==========================================
+ Coverage   90.66%   90.81%   +0.15%     
==========================================
  Files          12       12              
  Lines        2068     2102      +34     
==========================================
+ Hits         1875     1909      +34     
  Misses        193      193              

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.