The documentation of the source time function in terms of the hard-coded
function f'(t) on the one hand and our understanding of a source-time history
on the other hand was confusing. Hopefully this is resolved now.

When compiling sigfit, the overloaded function resolution algorithm selected
the wrong output operator for ts::TimSeries objects. An output operator for
writing instances of this object type to datrw::oanystream was missing.

For this reason the overloaded output operator for aff::Series type objects to
datrw::oanystream was selected by casting ts::TimeSeries to its public base
class type aff::Series.

Solved the problem by providing additional output and input operators in
libtsxx and access to the debug flag in ianystream and oanystream in
libdatrwxx.

provide input and output operators for ianystream and oanystream;
they are not inherited from idatstream and odatstream, type conversion
is not possible, when call the operators for TimeSeries objects

confirmed: program passes correct header date to libdatrwxx; file contains
wrong value

problem: sigfit produces unreasonable data start time values upon output of
residuals file

catch typo in SG056 channel configuration
Br2-F60 and B2r-F60 are the same data

Apart from primarily recorded channels the UIPC system provides signals
derived by signal processing. The comprise down-sampled versions of the
primary gravity and pressure recordings as well as synthetic tides and
residuals with respect to synthetics and between sensorѕ.

Move file from TFSoftware to Seitosh.

The calculation of the time of the current sample failed to handle time series
containers where the index of the first sample was finite. A finite first
sample index is a standard feature supported by libaff.

These filters effectively load a filter operator (be it IIR or FIR) with the
value of the first sample. This is done by first removing the value of the
first sample from the entire time series and by restoring this value after
application of the filter.

a new option for resaseda provides the shrink option in the resample function
of the interpolator

fix bug: negative values were not normalized correctly in cases where they
were integer multiples of the limit values; this cause problems, when calling
time_compare with the result from time_sub when the result of time_sub was not
properly normalized

Provide means to reduce the output time window to the time span for which
input data is available while maintaing the desired sampling raster. This is
useful for data with gaps.

We need a new exception to indicate an empty time window for cases, where the
interpolator shrinks the time window to fit into the time span of the input
time series.

  - automatically convert comma to whitespace in
    coordinate parameters
  - support creation of info data

Member data Mfile was initialized with numeric value 0. This was appropriate
when Mfile was of type char * which is no longer the case. Mfile now is of
type std::string and requires a character sequence as initializer.

be more explicit with respect to the effect of tshift and irtap. tshift just
affects the output impulse response and not the filter operation.

SRCE data is a file property not a trace property

The array is not just used for taking picks, but also for reading dispersion
relation data from files. When numerically sampling the curves with a small
interval, the number of samples may easily exceed the previous size od the
array dimensions.