Comparison of different esf files I found confirm this mystery byte after record type in ESF cannot be anything but version number.

Here are some records with multiple discovered version numbers, or with one but strangely high version number (format - [unique versions found, [vers, samples], [version, samples], :CODE]).
[1, [4, 383], :REGION]
[1, [4, 42], :FORT]
[1, [4, 673], :OBSTACLE]
[1, [4, 6], :CAMPAIGN_PREOPEN_MAP_INFO]
[1, [4, 6], :CAMPAIGN_SETUP_OPTIONS]
[1, [4, 6], :PATHFINDING_GRID]
[1, [4, 6], :WORLD]
[1, [5, 6], :CAMPAIGN_STARTPOS]
[1, [5, 6], :CAMPAIGN_TRADE_MANAGER]
[1, [6, 2168], :CHARACTER]
[1, [8, 6], :EPISODIC_RESTRICTIONS]
[2, [0, 18], [1, 365], :CAI_BDI_GOAL_REGION_DEFENCE]
[2, [0, 18], [1, 365], :CAI_DEFENCE_AND_INVASION_FORCE_STRENGTH_ANALYSIS_REGION]
[2, [0, 1], [1, 30], :CAI_BDIM_ATTACK_SH]
[2, [0, 3], [2, 105], :CAI_BDI_GOAL_ACQUIRE_REGION]
[2, [1, 2], [2, 4], :LOCOMOTION_MANAGER]
[2, [1, 2], [2, 65], :CAI_BDI_TARGET_REGION_BASE]
[2, [1, 35], [2, 351], :ARMY]
[2, [1, 3], [2, 65], :CAI_BDI_INVASION_REQUEST]
[2, [1, 9], [2, 134], :CAI_HIGH_LEVEL_PATH]
[2, [3, 2], [4, 4], :CAMPAIGN_CAMERA_MANAGER]
[2, [3, 7], [4, 132], :FACTION_TECHNOLOGY_MANAGER]
[2, [4, 2], [5, 4], :CAMPAIGN_PATHFINDER]
[2, [4, 6], [5, 167], :CAI_BDI_CORRIDOR]
[2, [4, 9], [5, 136], :ECONOMICS_DATA]
[2, [5, 2], [6, 4], :PENDING_BATTLE]
[2, [5, 9], [6, 136], :FACTION_ECONOMICS]
[2, [8, 22], [9, 5844], :DIPLOMACY_RELATIONSHIP]
[2, [8, 2], [9, 4], :CAMPAIGN_MODEL]
Version does not determine data stored - it's entirely usual for multiple possibilities to be available, different number of fields, usually prefixed by field count, or optional fields (<false /> vs <true /><blah />).

Nothing seems to imply that changing faction or region counts are hardcoded at this level - all this seems to be about features (I should try getting ancient .esfs just to be sure).