after days of ad9361 noise analysis with multiple borad, it turns out that the ad9361 RxFir can significantly reduce max noise level, with which the interference-to-noise ratio can be greatlt improved, and the nulling algotirhm can finally work properly; another experiment is about duifei multi rx channel phase-amplitude consistence, for which the rx external lo of ad9361 is enabled, but extra phase compensation is needed because the trace length of external lo is not matched
in dl814_k7_cal_coarse, change ant_clk to clk_96m, add rx_pre_test to verify nulling algorithm, with the ability to switch rx data between ad9361 data and test data
now rf control and lcmv autocorrelation data shares the same signal line by simply 'or' operation, so the two must be transmitted at different time, otherwise the data will collapse
josh