Fix Export Service Implementation Variable Initialization

The export service implementation class has global class variables that keeps state for the exporters, which poses a problem when multiple exports are done in immediate succession. That's why each exporter is now created within the function that utilizes it allowing for multiple calls to be made without worry of past state corrupting current exports.

Generalized stochastic simulations, improved VCell ImageJ Plugin

• Expanded rate laws for stochastic simulations. Previously VCell had only supported mass action rate laws for stochastic simulations. This Release allows any rate law expressed for an ODE simulation to be simulated stochastically.

• VCell ImageJ plugin has improved usability and performance for analyzing large VCell simulation datasets served remotely using the cloud ready N5 format. Data transfer is much faster and simulation data is resampled for improved ImageJ compatibility.

release candidate with generalized stochastic and imagej improvements

Merge pull request #1359 from virtualcell/use-vcell-fvsolver

simplified Slurm scripts, uses vcell-solvers and vcell-fvsolver containers

fixed vcell-fvsolver config in vcell-submit service

7.6.0.49

fixed config mistake for vcell-fvsolver name on vcell-submit dockerfile

improved Slurm submission script

fixed Slurm bash syntax
translated host to container paths for container arguments

dev release: remove singularity from CI/CD

singularity images are no longer prebuilt in CI/CD - find another way.

dev release: simplified Slurm scripts, multiple solver containers

7.6.0.45

Merge remote-tracking branch 'origin/master' into use-vcell-fvsolver

Include Z,Time, And Number of Variables in DTO's

Merge pull request #1358 from virtualcell/quick-n5-fix

Human Readable Passes Dimension Info to its Spec

Generalized Stochastic

nonspatial stochastic simulation (Gillespie/Gibson solver) now supports general kinetic laws.

• As before, reaction kinetics recognized as Mass Action are simulated using elementary forward and reverse stochastic processes.
• Now, non-mass action kinetics (e.g. enzyme kinetics) can be simulated stochastically where the propensity is derived from the effective net reaction rate.
  • Each reaction is still decomposed into forward and reverse stochastic processes,
  • but the forward rate propensity is driven by the net reaction rate (when its sign is positive),
  • and the reverse rate propensity is driven by the net reaction rate (when its sign is negative)
  • this approach results in reasonable mean trajectories, but may not capture rare events accurately and the noise characteristics of the underlying elementary processes (normally described by mass action kinetics) are not well represented.

Improve N5 Experience

Allow users to export post-processing variables in the N5 file format if that is the only type of variable available, and include dimensional size metadata when exporting N5 files.