The table in section F shows a system that approximates 606 synodic lunar months with 17895 days exactly. This gives an average synodic lunar month of 29.5297 days, which falls short of today's value, and falls even further short of 1st millennium BC values. It is still not clear to me how your system can make up the shortfall without violating one of its day-count constraints.
The article needs to be clearer about the system's rules. As I noted above, section F shows a jubilee of 17895 days. Nothing in the table or text allows for a jubilee to have any other number of days. Yet the table in section H suggests that possibly half of the jubilees have more days than this:
Sunset 29 March -904 to sunset 27 March -855: 17895 days.
Sunset 27 March -855 to sunset 26 March -806: 17896 days.
Sunset 26 March -806 to sunset 24 March -757: 17895 days.
Sunset 24 March -757 to sunset 22 March -708: 17896 days.
Sunset 22 March -708 to sunset 20 March -659: 17895 days.
Sunset 20 March -659 to sunset 20 March -610: 17897 days.
Sunset 20 March -610 to sunset 18 March -561: 17895 days.
If the 17718-day count is to be maintained, the extra days must always fall in the last 6 months of the jubilee. In the 17897-day jubilee this gives the unusual, though not impossible, situation of 6 consecutive lunar months only one of which has 29 days, the rest having 30. Apparently the last 1290-day count is not a strict rule, and this period can sometimes have 1291 or 1292 days. This makes up the lunar shortfall I mentioned earlier, but the article gives no rule for when these extra days should be added. Nor does it give a clear rule for when insertions of intercalary months will occur, merely stating somewhat vaguely that they will not always be in the order shown in section G.
The point of having a fixed lunar cycle rather than an empirical lunar calendar is to be able to know at the beginning of every year exactly how many months will be in the year, and exactly how many days will be in each month. There is no point in maintaining large day-counts over many years if the effort does not provide this convenience. But this means that the rules for intercalations and month-lengths need to be clear and well-known, so that computations can be done well in advance of the event.