Look, I think for the sake of this entire discussion, we should agree on some general principles. I would suggest:
1) Pumped hydro energy storage is practical and long-proven as a means of time shifting excess electrical energy. Indeed, it is fully mature and various hydropower dams have decades of operating experience. For those of you who do not know, Hoover dam has been in continuous operation since 1936 with an enviable safety record. Numerous engineering updates have been made including computer modeling of power grid ties, and computer modelling of long term power production. Lastly pumped hydro still comprises over 98% of the existing grid energy storage in 2018.
2) Construction of the pumps that pump uphill will utilize modern computer-aided design practices and losses in some similar and recently designed pumping stations in Norway are below 5%. This is about 1/3rd of the mechanical/electrical energy lost in the 1930's-era generators at Hoover dam, even after recent updates.
3) Performance of the dam and lake and pumping station would be maximized by having the lake full and the generators operating at 100% of capacity. At 100% head in terms of power generation, the dam and lake still retain about 35% extra capacity for flood abatement purposes, water would be buypassing the generators but the dam would not be in danger of breaching. Of course, as already noted, the water needed to fill the lake once to produce maximum head is not available in the area, this problem unless resolved would limit storage capacity and power production.
Source: IEEE Transactions on Power and Energy, which is unfortunately paywalled for non members. However I can provide additional details if needed.