Two standard OLS orthogonality identities: residuals sum to zero and are orthogonal to fitted values

Consider a multiple linear regression model $\underline y=X \underline B+\underline E_{n+1}$ and suppose $\hat y_i$ and $e_i\ i=1,\dots,n$ are the fitted values of response variable y and residuals respectively. Then: