300mS eliminated the hunting issue!

I upped the PWM sampling and adjustment period to 300mS, and that eliminated all hunting during the Acceptance phase:

Do note the spike around 11.5 minutes into the run, here something turned on (my guess is the freezer or refrigerator  looks like what ever it is had a surge current to start and a 8-10A DC steady state draw.  Sounds like the Refrigeration to me).

Because I was dumping debug data at every PWM adjustment period I could see exactingly how the system responded.  Here is a snip:

Time	mS	PWM	Vbat	Amps	Watts	RPMs	:Ve
11.2	718233	70	14.79	42	621	1834	0
11.3	718575	70	14.78	41	605	1842	0
11.3	718919	70	14.77	40	590	1834	0
11.3	719249	70	14.77	41	605	1827	0
11.3	719592	70	14.77	40	590	1842	0
11.3	719940	70	14.77	40	590	1842	0
11.3	720275	70	14.76	40	590	1834	0
11.3	720619	70	14.76	40	590	1827	0
11.3	720966	70	14.76	38	560	1842	0
11.3	721299	70	14.76	38	560	1834	0
11.3	721643	70	14.75	40	590	1834	0
11.3	721991	70	14.73	41	604	1842	2
11.3	722326	72	14.73	43	633	1834	2
11.3	722669	74	14.76	44	649	1849	0
11.3	723016	74	14.77	44	649	1820	0
11.3	723368	74	14.78	42	620	1820	0
11.3	723713	74	14.26	43	613	1834	16		VBat declining, looks like a load was placed on.
11.3	724057	76	13.94	45	627	1827	27		System responds by increasing PWM  (PWMs are capped max 2 going UP)
11.3	724384	78	13.95	44	613	1834	27		Still needs more PWM, but at least we have stopped the decline in VBat now.
11.4	724728	80	14.24	48	683	1827	17		Voltage is starting to rise.
11.4	725072	82	14.28	52	742	1834	16		However because we are still under the target voltage . .
11.4	725400	84	14.32	52	744	1812	14		. . simple algorithm continued to increase PWM
11.4	725744	86	14.36	56	803	1820	13
11.4	726094	88	14.41	56	807	1820	11		Still increasing, voltage is raising faster now.
11.4	726438	90	14.46	58	838	1820	10
11.4	726780	92	14.51	58	841	1805	8
11.4	727124	94	14.58	62	903	1812	6
11.4	727453	96	14.66	65	952	1790	3
11.4	727796	98	14.74	67	987	1790	2
11.4	728144	100	14.82	68	1008	1782	-1		Oh Oh, now we are over volts!  Need to bring down the PWM.
11.4	728488	99	14.91	68	1013	1775	-4		But we had wound it up so much on the way up it takes time to reduce.
11.4	728833	95	14.95	65	971	1790	-6		Notice on the way down no limit how much  we can change PWM in one step..
11.4	729177	89	14.96	60	897	1782	-6
11.4	729504	83	14.93	56	836	1797	-5
11.4	729846	78	14.88	51	758	1805	-3
11.4	730196	75	14.82	47	696	1827	-1
11.5	730531	74	14.76	45	664	1827	0
11.5	730874	74	14.72	45	662	1834	2		Only a little undershoot
11.5	731224	76	14.7	46	676	1827	2
11.5	731578	78	14.72	48	706	1842	2
11.5	731922	80	14.72	51	750	1827	2
11.5	732266	82	14.74	53	781	1820	0		Finally, back to stable VBat.  Amp draw is about 8-10A higher then before.
11.5	732621	82	14.76	49	723	1812	0
11.5	732965	82	14.76	52	767	1827	0
11.5	733318	82	14.78	50	739	1820	0

When I look at this, again I clearly see how adding in consideration for Trending will help.  And I am likely going to make that change over the summer.  But for now, I think this will work OK.  I am going to try lower the cap to a max change of 1 to try and reduce overshoot.  And I worry about the systems ability to respond to over-voltage, so am going to try and reduce the sampling time to 200mS and see if that still mitigates the hunting conditions.  I might even be able to get away with 150mS, when combined with reducing the CAP from 2 to 1.