GitLab

- use forward declaration where possible
- remove unused includes

cleanup: merge the duplicate code of LF2 and RK4into a function initDistInGlobleFrame(), not used in MTS tracker since units are different

fix occasional crash after particles is deleted by the plugin elements in cyclron tracker, destroy particle in the local frame

fix the mass in *.loss file with more than one plugin element case (open/close the file just before / immediately after data dump)

    - deleting obsolete find modules
    - using HINT and adding default library and include path
      (CPP_INCLUDE_PATH and LIBRARY_PATH) in GSL, H5hut and HDF5 find modules

Set Option REMOTEPARTDEL to Artifically delete the remote particle if its distance to the beam mass is larger than this factor times the beam rms size, its default values is -1 (no delete), current only valid for OPAL-cycl

- Dump the info of stripped particle to the shared *.loss file, rather than an separated *.extract file; \n - Dump the turn number and time for the probed or stripped particle

    - Different splitting allows any positive integer for MTSSUBSTEPS option
    - Requires still one space charge solve per step, but one additional
      solve before main loop. Therefore, extract code to function
      evaluateSpaceCharge
    - TODO: correct location of bunch injection code?

coding guidelines. (More to do here.)

	- applyPluginElements expects [R] = mm, [dt] = ns
	- but Tracker_MTS uses [R] = m, [dt] = s

heighten the doorsill for artificially deletion of the romate particles to 8*sigma in OPAL-CYCL, which is the case of quite unmatched initial beam distribution

Merge branch 'develop' into svn: fixing long standing issue with number of steps performed with track

Merge branch 'develop' into svn: get rid of half of the remaining warnings in opal (without pragmas), disable the other half with pragmas;
 - move saving of maxSteps_m inside autophasing method
 - when using mpicxx not with its default compiler then the cmake file should still work. extract the absolute path to the underlying compiler and its name

Workaround, because doAutophase is changing the maximum number of steps we defined in the TRACK cmd.

Merge branch 'develop' into svn: unify syntax; adding default CMAKE_BUILD_TYPE; put PROJECT(OPAL) to the top

If the query

    bool isOPALt();

if not isOPALt() we assume it is OPAL-cycl

The following two functions should probable be private
and called when the h5 file is openend.

    void setOPALcycl();

    void setOPALt();

This is a first implementation to get Jianjun going
with the OPAL-t OPAL-cycl 

Merge branch 'develop' into svn: fixing autophase by merging the correct traveling wave list into cavities_m

Merge branch 'develop' into svn: exporting the new c++11 disable flags (nullptr and foreach) as cmake options