We verify that, in a linear model, the likelihood value obtained using
the SIS algorithm is equal to the likelihood value calculated with a
standard linear Kalman filter.

Related to #1935

Also add a test with dseries but with bare shocks/endval.

- Remove globals
- Remove unused code paths
- support steady_state_model
- support contemporaneous_correlation
- support TeX option

added Irreversible_investment_leadlag.mod to testsuite and make comparison with Irreversible_investment.mod

[skip ci]

Fixes https://forum.dynare.org/t/invalid-mex-file-whe-running-sensitivity-analysis-m1-mac-6-1/25829

It seems that meson does not link libquadmath (which is needed by libgfortran).
When doing dynamic linking, this is not an issue since the dynamic linker loads
libquadmath recursively after loading libgfortran. But when doing static linking,
libquadmath needs to be explicitly linked.

Not yet clear at this point: why isn’t this needed under Linux and Windows?

Under Debian 12, it fails with:
/usr/bin/ld: /usr/lib/gcc/x86_64-linux-gnu/12/libgomp.a(team.o): réadressage R_X86_64_TPOFF32 vers symbole caché « gomp_tls_data » ne peut pas être utilisé en créant un objet partagé

xvfb-run is a Debian-specific program.

Closes: #1892

Dynare++ is no longer distributed as a standalone binary.

Fix marginal linearization in the context of perfect_foresight_with_expectation_errors_solver with homotopy

If a solution corresponding to 100% of the shock can’t be found in the first
informational period, marginal linearization will be performed to extrapolate a
solution.

However, in subsequent informational periods, this extrapolated solution cannot
be used for the initial conditions of endogenous variables, because the initial
conditions are not a true solution of the nonlinear model. For those subsequent
informational periods, the correct approach is to compute the two solutions
needed for marginal linearization using as initial conditions the values
obtained in the same two solutions for the previous informational
periods (stored as oo_.deterministic_simulation.{sim1,sim2}).

Unfortunately it is not possible to enable it for all supported Octave
versions (which would have allowed us to remove the old codepath), because of
this bug: https://savannah.gnu.org/bugs/?64687

Incidentally, this means that the JSONlab submodule is no longer needed, so it
is removed in the present commit.

Closes: #1907

Ref. #1907