Fix: Update Mitochondrial reaction reversibility and localization

[hchapman1]

Main improvements in this PR:

Summary
This PR implements mitochondrial-focused updates on feat/curateMitochondria, addressing discrepancies between Human-GEM and MitoCore, a highly curated model of human central metabolism. The following changes were made:

• Updated reversibility (bounds + rev) for 38 reactions (listed in issue Update mitochondrial reaction bounds based on Gibbs free energy and MitoCore comparison #1006).
• Duplicated 6 reactions into an additional compartment (mitochondrial or cytosolic), based on agreement with MitoCore and mitochondrial localization evidence from MitoCarta.

For reversibility mismatches between models:
Reversibility was investigated for reactions where Human-GEM and MitoCore disagreed. For these reactions, I queried the standard Gibbs free energy (ΔG°′) using eQuilibrator and applied the following rule to decide reaction directionality:

• If −30 kJ/mol ≤ ΔG°′ ≤ +30 kJ/mol → reaction treated as reversible
• If ΔG°′ < −30 kJ/mol → reaction constrained to be forward-only (lb = 0, ub = 1000)

See the full list of bounds changed in issue #1006.

For compartment mismatches between models:
I identified 17 reactions with mismatching compartments between HumanGEM and MitoCore. Six of these were resolved by duplicating reactions across compartments (rather than moving them) to reflect mitochondrial localization supported by MitoCore and independent localization evidence. For reactions duplicated into the mitochondrial compartment, the associated proteins have mitochondrial localization evidence in MitoCarta 3.0.

HumanGEMID	current_comp	duplicated_to	Duplicate ID
MAR03802	m	c	MAR020189
MAR03804	m	c	MAR020190
MAR04149	c	m	MAR020193
MAR04336	c	m	MAR020194
MAR06518	c	m	MAR020192
MAR08512	c	m	MAR020191

The remaining 11 compartment mismatches were kept in their original Human-GEM compartment, either because the required metabolites were not available in the proposed compartment in Human-GEM or because the available evidence was insufficient to justify reassignment or duplication.

Testing
Model was tested using Raven's checkTasks on the following metabolic task sets: Essential Tasks, Verification Tasks and Full task list.

I hereby confirm that I have:

• Tested my code on my own computer for running the model
• Selected develop as a target branch
• Any removed reactions and metabolites have been moved to the corresponding deprecated identifier lists

[github-actions]

This PR has been automatically tested with GH Actions. Here is the output of the MACAW test:

Starting dead-end test...
 - Found 1381 dead-end metabolites.
 - Found 1137 reactions incapable of sustaining steady-state fluxes in either direction due to these dead-ends.
 - Found 1367 reversible reactions that can only carry steady-state fluxes in a single direction due to dead-ends.
Starting duplicate test...
 - Skipping redox duplicates because no redox_pairs and/or proton_ids were provided.
 - Found 379 reactions that were some type of duplicate:
   - 0 were completely identical to at least one other reaction.
   - 13 involve the same metabolites but go in the opposite direction or have the opposite reversibility as at least one other reaction.
   - 379 involve the same metabolites but with different coefficients as at least one other reaction.

This and a more detailed output from MACAW are also committed to data/macawResults/.

Note: In the case of multiple test runs, this post will be edited.