Sundials solver is allowed to warn and continue if it returns a timestep
with no change in t.
If the solver doesn't recover, it will fail with eg SciMLBase.ReturnCode.MaxIters

Reverts PR #416, fixes
#420

For CVode and ARKode, it looks like this is intended behaviour: the
Sundial solver emits a warning message, with an API call
    CVodeSetMaxHnilWarns
    ARKStepSetMaxHnilWarns
to set the maximum number of warning messages printed, which is exposed
to Julia as
    max_hnil_warns
see Sundials driver code https://github.com/LLNL/sundials/blob/e8a3e67e3883bc316c48bc534ee08319a5e8c620/src/cvode/cvode.c#L1339-L1347

For IDA, there is no API like this so it's not so clear what should happen,
however the behaviour prior to #416 (restored by this PR)
for the f_noconverge test added in test/common_interface/ida.jl is to return
  SciMLBase.ReturnCode.MaxIters
which seems reasonable?

(NB: @oscardssmith I can't reproduce the issue implied by the title of
#416 so perhaps what is missing here
is the original failure case that demonstrates the issue?
The call to solve in the f_noconverge test test/common_interface/ida.jl, returns SciMLBase.ReturnCode.MaxIters,
it doesn't return with no error?

Also SciML/SciMLBase.jl#458 shows IDA printing
  [IDAS ERROR]  IDASolve
    At t = 0 and h = 1.49012e-18, the error test failed repeatedly or with |h| = hmin.
which suggests that IDA did flag an error in that case ?
)

Bumps [actions/cache](https://github.com/actions/cache) from 3 to 4.
- [Release notes](https://github.com/actions/cache/releases)
- [Changelog](https://github.com/actions/cache/blob/main/RELEASES.md)
- [Commits](actions/cache@v3...v4)

---
updated-dependencies:
- dependency-name: actions/cache
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <[email protected]>

…tstops`

this version works when tstops is a scalar.

Addresses #435 and
also fixes a potential problem with memory safety:

1) Allocations in convert(Ptr{T}, Handle{T})
   Fix is as suggested in #435
   (Handle is now a mutable struct with a ptr field)

2) Memory safety robustness fix
   Remove convert and use paired Base.cconvert / Base.unsafe_convert to get the ptr field from
   a Handle object h, so that the h is preserved from GC across the ccall
   This fix is analogous to that for the NVector wrapper in PR #380
   (NB: there is no actual problem at least when Sundials is used with the SciML interface,
   as the Handle{CVODEMem} and similar objects are held by a persistent solver data structure, but this
   change should reduce the risk of something going wrong in the future, or for eg test harnesses that don't use
   the SciML interface)