`flux_ranocha` is not entropy conservative for `CompressibleEulerMulticomponentEquations1D`

The implementation of `flux_ranocha` for `CompressibleEulerMulticomponentEquations1D` is not entropy conservative when `gammas` or `gas_constants` are not identical. `flux_chandrashekar` seems OK though.

Additionally, the `initial_condition_weak_blast_wave` used in `examples/tree_1d_dgsem/elixir_eulermulti_ec.jl` results in `rho1`, `rho2`, `rho3` which are identical up to a constant scaling - this also seems to preserve EC up to machine precision.

Here is a MWE where `rho_i` are not scalings of each other and `gammas`/`gas_constants` are not identical.
```julia
using OrdinaryDiffEqLowStorageRK
using Trixi

equations = CompressibleEulerMulticomponentEquations1D(gammas = (1.4, 5/3, 4/3),
gas_constants = (0.4, 0.6, 0.5))

function initial_condition_random(x, t, equations)
rho1, rho2, rho3 = ntuple(i -> 1 + 0.1 * sin(i + x[1]), 3)
v1 = 0.1
p = 1 + 0.1 * sin(x[1])
q = SVector(v1, p, rho1, rho2, rho3)
return prim2cons(q, equations)
end
initial_condition = initial_condition_random

volume_flux = flux_ranocha
solver = DGSEM(polydeg = 3, surface_flux = flux_ranocha,
volume_integral = VolumeIntegralFluxDifferencing(volume_flux))

coordinates_min = (-2.0,)
coordinates_max = (2.0,)
mesh = TreeMesh(coordinates_min, coordinates_max,
initial_refinement_level = 4,
n_cells_max = 10_000, periodicity = true)

semi = SemidiscretizationHyperbolic(mesh, equations, initial_condition, solver;
boundary_conditions = boundary_condition_periodic)

tspan = (0.0, 1e-2)
ode = semidiscretize(semi, tspan)

analysis_interval = 10
analysis_callback = AnalysisCallback(semi, interval = analysis_interval,
extra_analysis_integrals = (Trixi.density,))

alive_callback = AliveCallback(analysis_interval = analysis_interval)

stepsize_callback = StepsizeCallback(cfl = 0.1)

callbacks = CallbackSet(analysis_callback, alive_callback,
stepsize_callback)

###############################################################################
# run the simulation

sol = solve(ode, CarpenterKennedy2N54(williamson_condition = false);
dt = 1, # solve needs some value here but it will be overwritten by the stepsize_callback
ode_default_options()..., callback = callbacks);
```
The analysis output yields ` ∑∂S/∂U ⋅ Uₜ : -6.80429623e-06`

Start by reproducing the supplied MWE for CompressibleEulerMulticomponentEquations1D with unequal gammas, gas_constants, and non-scaled rho_i. Read the flux_ranocha implementation and compare its entropy behavior with flux_chandrashekar; the example is examples/tree_1d_dgsem/elixir_eulermulti_ec.jl. Done means the entropy residual no longer shows the reported violation for the MWE.

Written by the indexing model from the issue text.