Mitochondrial fatty-acid synthesis is in the cytosol, so lipoic acid synthesis is disconnected #1102

Mitochondrial fatty-acid synthesis (mtFAS) makes octanoyl-[ACP] in the mitochondria, the starting point of lipoic acid synthesis (LIPT2 → LIAS). In Human-GEM this pathway cannot run:

• Wrong compartment: the elongation reactions MAR02151–MAR02173 (malonyl-[ACP] up to 3-oxotetradecanoyl-[ACP]) are in the cytosol and use [ACP][c], although their GPRs are the mitochondrial enzymes MCAT, OXSM and MECR.
• Only one step in the mitochondria: MAR01517 makes malonyl-[ACP][m] from [ACP][m] (MAM00184m). That is a different ACP species from the mitoACP (MAM02484m) that lipoylation uses, and MAM00184m occurs in no other reaction.
• No source for octanoyl-[ACP][m]: MAM02643m is only consumed, by MAR06402. So de novo lipoylation (MAR06402 → MAR06403 → MAR06404 → MAR20168) is a dead end, and mitochondrial lipoamide can only come from import (MAR11361).

Proposed:

• Build mtFAS in the mitochondria on mitoACP, from malonyl-[ACP] to octanoyl-[ACP].
• Enzymes: MCAT; OXSM (condensation); CBR4 with HSD17B8 (3-ketoacyl reduction); HTD2 (dehydration, not yet in the model); MECR (enoyl reduction).
• Merge or remove the separate [ACP][m] species.
• Check whether the cytosolic copies are correct. Cytosolic FASN also makes acyl-ACPs, but not with these GPRs.

This removes MAR06402 and MAR20168 from the MACAW dead ends (see #1099).

Also covers #684. #684 lists the metabolites, the reactions, and their GPRs and references for the stepwise pathway from acetyl-/malonyl-ACP to octanoyl-ACP, including the new gene HTD2 (ENSG00000255154). Build the pathway as proposed there, on mitoACP (MAM02484m), and remove the separate [ACP][m] (MAM00184m).

Also check why growth needs mitochondrial apo-ACP import and breakdown (MAR05222, MAR05267; see the comment below).