Please find below the responses to your questions.
Could you clarify please, whether sum of the potential losses is equal to the amount of instantaneous loss defined in Article 121 (9)?
EIOPA: The ex-post adjustment can be applied only to the nat cat module for the EEA countries (See "EIOPA's second set of advice to the European Commission on specific items in the Solvency II Delegated Regulation" 6.5. Contractual limits and natural catastrophe risk). It therefore can't be applied to the instantaneous loss as defined in Article 121 (9).
As per the first question, I could not find the definition for potential loss within the document, apart from the formula for instantaneous loss.
EIOPA: The potential losses refer to the weighted sum insured "WSI" as defined in Article 121 (b) subparagraph (i). The steps described p.123 in "EIOPA's second set of advice to the European Commission on specific items in the Solvency II Delegated Regulation" found on EIOPA's website might help to understand how the ex-post adjustment is applied.
2. Article 90b "Simplified calculation of the sum insured for natural catastrophe risks", states that 'the risk weight for windstorm risk referred to in point (a) of Article 121(6) shall be the risk weight for windstorm risk in the risk zone within that group with the highest risk weight for windstorm risk'. Can you clarify please, how this will affect the correlations used in the windstorm loss formula? Do we assume that the correlation will equal to 1?
Regarding the second question; when simplification is applicable with risk weight being the highest risk weight in the risk zone within the group, I think this leads to the correlation coefficient for windstorm risk in risk zones i and j of region r set out in Annex XXII in windstorm loss formula becoming one, due to the fact that only one risk zone per country is considered.
EIOPA: Indeed, if we assume only one risk zone per country, the correlation coefficient for windstorm risk in risk zones i and j of region set out in Annex XXII becomes useless. Please find below a few examples which show how the simplification can be used:
If we assume in a simple example that Q(Peril, Country) =1 and a country with three CRESTA zone:
Z1 / risk weight = 0.1
Z2 / risk weight = 0.8
Z3 / risk weight = 0.3
We also assume the following Correlation matrix:
    Z1    Z2    Z3
Z1    1    0.5    0.3
Z2    0.5    1    0.9
Z3    0.3    0.9    1

Example 1:
We know there is 1 Mio. Sum insured in each zone.
Z1 / risk weight = 0.1 / WSI1=0.1
Z2 / risk weight = 0.8 / WSI2=0.8
Z3 / risk weight = 0.3 / WSI3 = 0.3
In case 1, L(peril,country) = sqrt(WS1^2+WS2^2+WS3^2+2*0.5*WSI1*WS2+2*0.3*WS1*WS3+2*0.9*WS2*WS3)=1.13
Example 2:
We need to use the simplification as we know that 1 Mio. sum insured can be assigned to zone 1 but we are not sure about the other 2 Mio. We therefore assign the 2 Mio to the zone which has the highest risk weight (i.e. in our example Zone 2).
Z1 / risk weight = 0.1 / WSI1=0.1
Z2 / risk weight = 0.8 / WSI2=1.6
Z3 / risk weight = 0.3 / WSI3 = 0
In case 2, L(peril,country) = sqrt(WS1^2+WS2^2+2*0.5*WSI1*WS2)=1.65
Example 3:
We don't know to which zone we can assign the 3 Mio. sum insured.
Z1 / risk weight = 0.1 / WSI1=0
Z2 / risk weight = 0.8 / WSI2=2.4
Z3 / risk weight = 0.3 / WSI3 = 0
In case 3, L(peril,country) = WS2 = 2.4 as the entire sum insured is assigned to one zone (with highest risk weight) the correlations are not needed anymore.
3. In addition, it would be greatly appreciated if you could specify when these amendments are expected to be inforce.
EIOPA: The amendments will be in force from the 8th of July 2019.