welcome censored data into loadflex

[aappling-usgs]

would like to talk with you, @ldecicco, about what the basic loadflex format for censored data should be. we have the option to make it something other than smwrQW, i think, as long as we can translate it into/out of that format for loadReg2 calls.

regardless of the loadflex format, we want to be able to transfer censored data over to EGRET format and therefore need to have a plan for how those data will be formatted in loadflex. this issue is therefore related to #93

[aappling-usgs]

EGRET uses ConcLow, ConcHigh, Uncen, and ConcAve. We could use something similar. From https://pubs.usgs.gov/tm/04/a10/pdf/tm4A10.pdf page 8 (17/104):

The concentration information is stored using two variables, one called ConcLow
and the other called ConcHigh. When the data are uncensored, these two variables are equal to each other (set to the reported concentration value). If a value is censored, for example with a reporting limit of 0.5 mg/L, then ConcLow is set to NA (which stands for Not Available) and ConcHigh is set at the reporting limit, in this case 0.5. The variable called Uncen is always 1 for an uncensored value and 0 for a censored one. The mean value of Uncen for any given data set is equal to the frequency of uncensored values in the data set; for example, if the mean of Uncen is 0.95, then 95 percent of the values are uncensored and thus 5 percent of the values are censored. The variable ConcAve is provided for convenience and is used in some of the graphics functions but is not used in any computation. For censored values where ConcLow is equal to NA, ConcAve it is set equal to 0.5*ConcHigh. This type of censoring is referred to in the statistical literature as “left censoring.”